There's a scene early in Netflix's Adolescence where Detective Inspector Luke Bascombe listens to a voicemail from his teenage son. Adam's claiming to be sick, asking to skip school. Bascombe chuckles at the obvious lie, tells his partner his wife will handle it.

Minutes later, he's leading armed police through another family's door to arrest a 13-year-old for murder.

The boy, Jamie Miller, stabbed his classmate Katie seven times. Spent months in his bedroom consuming manosphere content that told him girls like Katie deserved violence. His parents were downstairs the whole time.

Adolescence became the most-watched streaming series in UK history. Keir Starmer made it free for schools. Everyone's asking the same question: how does a 13-year-old become capable of this?

But I think the show is actually asking something else. Something quieter and more uncomfortable. Why didn't any adult notice until police kicked the door down?

The detective who couldn't decode his own son's world

By Episode 2, Bascombe's at Jamie's school trying to understand motive. He's examining Instagram comments between Jamie and Katie; emoji exchanges that look like normal teenage social media stuff.

Then his son Adam pulls him aside. "You're not getting it. You're not reading what they're doing."

Adam explains that Katie wasn't flirting with Jamie. She was calling him an incel. Those emoji combinations? Red pill references. Manosphere code. The 80/20 rule that claims most women only want a small percentage of men.

Bascombe asks: "Have you been watching The Matrix?"

Think about this for a second. A detective—someone trained to gather information, understand motivation, read between the lines—doesn't know the manosphere exists. Can't decode interactions his own son navigates every day.

Adam had the information that could crack the case open. He just didn't think his father would understand it.

Struggling children use platforms differently—and you only will notice if you have been paying attention to them

A few weeks ago, researchers at the University of Manchester published findings from a three-year study tracking 25,000 students. They were looking for the causal link between screen time and mental health deterioration.

They didn't find one.

What they found instead: teenagers who are already struggling use social media differently than their peers. They use it more intensely. For mood regulation. To fill gaps that exist whether the platforms exist or not. The platforms aren't making them struggle. The platforms are what they reach for when they're struggling.

Here's the part that matters: you can only see these "different usage patterns" if you know your child well enough to notice the change.

The study accidentally confirmed something we don't like admitting. The question isn't "how many hours is your child online." It's "do you know your child well enough to notice when something's wrong."

What we've normalised

Adolescence keeps showing you the same pattern. Adults who don't know what children are experiencing until it explodes into crisis.

Jamie's parents knew he was in his room a lot. They knew he wasn't into sports. They didn't know he was spending hours consuming content teaching him that violence against women was justified. Bascombe knew his son was on his phone. He didn't know Adam was navigating manosphere culture at school every day.

We've started treating teenage withdrawal as natural. Developmentally appropriate. Something to shrug at rather than investigate.

"Teenagers need privacy."

"They're pulling away—it's normal at this age."

"I can't force them to talk to me."

All true. And also—conveniently—all reasons not to do the uncomfortable work of staying connected to a child who's making it difficult.

What compounds in the gap

Every online danger we worry about gets exponentially worse when children don't have a trusted adult they can turn to.

Grooming works because predators offer the attention the victim craves. They ask how the child's day was. They remember details. They make the child feel seen in ways their actual family doesn't. AI chatbots become confidants because they don't interrupt, don't dismiss, don't change the subject to talk about themselves. They're available at 2am when intrusive thoughts spiral.

Deepfake bullying goes unreported because the child already knows: there's no point bringing this to adults. Sextortion attempts that start with "I have screenshots that will ruin your life" only work because the teenager genuinely believes their parents would be angry rather than protective.

Platforms aren't creating these gaps. But they're very, very good at filling them when parents aren't.

The policy response that misses the point entirely

After Adolescence became unavoidable, Keir Starmer made it free for schools to screen. Teachers are being instructed to facilitate discussions about online radicalisation, toxic masculinity, the manosphere.

Let me be very clear: there's value in this. Schools need to address these topics.

But here's what's quietly absurd about the response. The show depicts a failure of adult attention at every level. Parents who didn't notice months of concerning behavior. Teachers who didn't intervene in visible social dynamics. Systems that only activated when a girl was dead.

And the institutional response is... to screen the show in schools. To put the burden on educators who are already drowning.

Where's the version where families watch Episode 2 together and then have the deeply uncomfortable conversation about whether the child could come to them if they were struggling? Where's the policy that says: "Before we ask teachers to fix this, let's ask parents if they know what apps their children use. What their online social life looks like. Whether their child believes they can come to them when something goes wrong."

Two fathers, two endings

Episode 4 of Adolescence takes place 13 months after Jamie's arrest. It's Eddie Miller's 50th birthday—Jamie's father.

The family is trying to maintain some version of normal. Birthday breakfast. Errands. Plans for Chinese food later.

Then Jamie calls from prison. He's decided to plead guilty. No explanation, just the announcement.

Eddie and his wife Manda have a breakdown in their kitchen. Asking each other where they went wrong. Should they have bought him the computer? Should they have monitored it more? How did they "make" a murderer when their daughter turned out fine?

The final scene is Eddie alone in Jamie's bedroom. The room hasn't been touched since the arrest. This is where Jamie spent all those hours. This is where the person he became was created, while his parents were downstairs.

Eddie sits on the bed. Picks up Jamie's teddy bear. Tucks it in like it's his son. Apologises to the stuffed animal because his actual son is in prison and unreachable.

His last words: "I should have done better."

The show never tells you what "better" would have looked like. There's no moment where you think "ah, if only they'd done X." Because there isn't one moment. There are a thousand small moments of not being present. Not listening. Not noticing.

Here's what makes Adolescence devastating rather than just depressing: Episode 2 ends differently for Bascombe and Adam.

After Adam explains the manosphere culture his father doesn't understand, after Bascombe realizes how little he knows about his son's daily reality, there's a shift. Bascombe asks Adam to get food together. Says "I love you. I want to spend more time with you."

Adam softens. They go to the Chinese restaurant.

The camera pulls back, showing this moment of repair, this father and son sitting across from each other with actual connection. Then it pans across the street to Eddie Miller placing flowers at Katie's memorial.

One father can still salvage his relationship. The other is trying to atone for what can't be undone.

What that moment of repair actually looked like

Think about what Bascombe did. He didn't lecture Adam about staying safe online. He didn't implement new rules or restrictions. He didn't install monitoring software or demand phone access.

He admitted he didn't understand his son's world. He asked to spend time together. He said the quiet part out loud: "I want to spend more time with you."

That's it. That's what noticing looks like. Not grand interventions. Just recognising the gap and taking a step toward closing it.

Adam had been lying about being sick that morning. The lie felt safer than the truth. But when his father showed genuine interest—not judgment, not control, just interest—Adam responded.

The change wasn't in Adam. It was in whether he believed his father would understand if he tried.

What actually protects children online

Every platform we're worried about. Every algorithm optimising for engagement. Every predator looking for vulnerable targets. Every piece of harmful content one search away.

All of it gets filtered through a single question when it matters: Does this child believe they can come to a trusted adult?

Not "are adults available"—most parents would say yes to that. Does the child believe it? Do they think telling you will help, or will it make things worse?

The Manchester study is showing us that struggling kids use platforms for mood regulation. They're already struggling—the platforms just fill the gap. The question isn't whether your child uses TikTok. It's whether you'd notice if their usage pattern changed. If they suddenly needed it more. If it was filling a gap that didn't exist before.

You can only notice that if you know your child well enough. If the connection is strong enough that changes are visible.

The uncomfortable truth

I know this is hard to hear. We want the solution to be technical. Install this app. Block this content. Limit screen time to X hours.

Because those solutions don't require us to examine whether our children trust us. Whether we're actually available when they try to talk. Whether we interrupt, dismiss, or change the subject when they bring up something we don't understand.

Bascombe didn't know what "red pill" meant. Instead of admitting that, instead of asking Adam to explain, his instinct was to reference The Matrix—to stay in his own frame of reference rather than enter his son's. How often do we do that? Child mentions something from their online world, and we either dismiss it as "internet stuff" or lecture about dangers without actually understanding what they're describing? I am guilty of it myself.

Adam lied about being sick because he didn't want to go to school that day. The lie felt safer than the truth. That's the signal. Not what he was avoiding—but that lying felt less risky than honesty.

What this means

I'm not saying connection solves everything. Jamie's parents might have been perfectly attentive and warm, and he still could have found that content, still could have been vulnerable to it.

The collective problems remain. Platforms are designed to be addictive. Algorithms amplify extreme content. Predators exist. Age verification doesn't work. Tech companies lobby against regulation. We need policy solutions. We need platforms to take responsibility. We need better digital literacy education.

But here's what's true regardless: If your child doesn't believe they can come to you, none of the other protections work.

The monitoring app doesn't help if they won't tell you why they're upset. The screen time limit doesn't help if you don't know what they're struggling with. The school program doesn't help if home is where they learned that adults don't listen.

The question the show is asking

Adolescence keeps returning to this image: Jamie's bedroom. Where he spent all those hours while his parents were in the house.

Were they neglectful? The show doesn't say that. Eddie worked hard, provided for his family, tried to do better than his own father did. But somewhere along the way, Jamie stopped believing his parents would understand. Stopped thinking it was worth trying to explain.

The show ends with Eddie wishing he'd spent more time in that room. Bascombe's story ends with him sitting across from his son at a restaurant, trying to bridge the gap before it becomes unbridgeable.

The difference between these two fathers isn't love. It's not even attention, exactly. It's whether they noticed the disconnection in time to do something about it.

Here's what the show doesn't tell you, but every parent knows: maintaining connection with a teenager who's actively pushing you away is brutally difficult.

They don't want to talk. They roll their eyes when you ask about their day. They retreat to their rooms and communicate in grunts. Bascombe's moment with Adam looks simple on screen—just ask to spend time together, say you love them. But that only works if you haven't already had fifty conversations where they've made it clear they don't want your attention.

What if you're already where Eddie was? What if the gap is already wide and you don't know how to cross it? What if you try to connect and your teenager shuts down harder?

I don't have a formula for that. Neither does the show. What Adolescence is showing us is starker: by the time you're asking "how do I reconnect?", you've already lost ground you might not get back.

That's terrifying. It should be.

But here's what's also true: teenagers are wired to push away from parents. It's developmentally normal. The difference isn't whether they push—it's whether they believe you'll still be there when they need you. Whether the door stays open even when they're slamming it.

Bascombe's son was lying to him that morning. Pushing away. Keeping secrets. And when Bascombe noticed—not with judgment, not with control, just with "I want to know you better"—Adam responded.

Not immediately. Not enthusiastically. But he went to dinner with his father.

That's what repair looks like. Not perfect communication. Not instant trust. Just a small step toward connection when you've noticed it fraying.

What you can do

I said this piece wouldn't give you a checklist, and I meant it. There's no "five steps to digital parenting" that solves this.

But if you're reading this and feeling the uncomfortable recognition that you don't actually know what your child does online, or you can't remember the last real conversation you had, or your teenager has stopped trying to explain things to you, you probably already know what the first step is.

It's not implementing new restrictions. It's not having a big serious talk about online safety. It's not demanding access to their phone.

It's what Bascombe did. Admitting you don't understand their world. Asking to spend time together—real time, not performative family dinner where everyone's on their phones. Saying out loud that you want to know them better.

Your teenager might say no. They might roll their eyes. They might agree to dinner and spend it in monosyllables.

Do it anyway.

Because here's the thing the Manchester study and Adolescence are both showing us: the danger doesn’t start with platforms. The danger is the gap the platforms fill. And the only person who can close that gap is you.

Not perfectly. Not all at once. But consistently enough that when your child is struggling—when they encounter something online that scares them, confuses them, or makes them feel unsafe—their first instinct is to come to you rather than hide it.

That instinct is built in a thousand small moments. In whether you listen when they try to explain something you don't understand. In whether you stay curious instead of immediately worried. In whether they believe you're on their side.

Eddie Miller sits in his son's bedroom wishing he'd built those moments. Bascombe caught himself before it was too late.

The show is asking: which parent are you going to be?

Not which parent are you now. Which parent are you going to be tomorrow, when your child mentions something from their online world and you have the choice to dismiss it or lean in. When they lie about something small and you have the choice to punish the lie or ask why the truth felt too risky.

Those moments are still happening. The choice is still yours.

Your eight-year-old asks to spend £4.99 on "gems" in their favourite game. The request seems reasonable—it's their pocket money, after all. What you don't see is the deliberate design architecture that led to this moment: the countdown timer creating artificial urgency, the virtual currency obscuring real costs, the social pressure from other players' purchases, and the reward mechanism calibrated to trigger one more ask, then another.

This is far from accidental. It's engineered manipulation, and it has a name: dark patterns.

What Makes a Pattern "Dark"

Dark patterns are interface design choices that benefit the company at the user's expense. Originally coined by designer Harry Brignull in 2010 to describe e-commerce tricks, the concept has evolved as researchers document how these techniques permeate children's digital experiences—from educational apps used in classrooms to the gaming platforms where they spend hours unsupervised.

Unlike straightforward deception, dark patterns exploit the gap between how interfaces appear and how they actually function. A 2024 study analysing apps used by preschoolers aged 3-5 found manipulative design features in 95% of free apps, compared to 80% of paid apps.

The difference isn't about honesty versus dishonesty—it's about understanding the sophisticated psychological mechanisms these designs trigger, particularly in developing brains.

Consider the temporal dark pattern called "playing by appointment." Games require players to return at specific times to collect rewards or complete tasks. Miss the window, lose progress. For adults, this might be annoying. For children still developing executive function—the cognitive ability to plan, organise, and resist impulses—it creates a compulsion loop their brains aren't equipped to manage.

Why Children Are Structurally Vulnerable

A German study published in 2024 tested 66 fifth-graders (ages 10-11) on their ability to recognise dark patterns. When explicitly asked to search for manipulations, about half spotted overly complex wordings and colour-based tricks. Only 1 in 4 identified manipulative formulations designed to steer decisions.

The children understood they were being influenced. They described feeling "tricked" even when they couldn't articulate exactly how. Yet understanding manipulation doesn't equal resistance—especially when design exploits developmental vulnerabilities that exist regardless of awareness.

Dr. Jenny Radesky, a pediatrician studying manipulative design in children's media, identifies specific cognitive limitations that dark patterns target:

Immature executive functions. Children are still developing self-control and decision-making capabilities. They struggle to resist visual triggers adults might dismiss: sparkles, countdown clocks, cascading rewards. The prefrontal cortex—responsible for evaluating consequences and inhibiting impulses—doesn't fully mature until the mid-20s.

Reward susceptibility. Children's behaviour is powerfully shaped by positive reinforcement. Dark patterns exploit this through variable reward schedules (the same mechanism behind slot machines), streak systems that punish missed days, and achievement frameworks that feel like progress but primarily serve engagement metrics.

Concrete versus abstract thinking. Children struggle to grasp the abstract scale of algorithms and data harvesting. They're cautious about stranger danger but don't transfer that caution to algorithms curating their feeds or companies profiling their preferences. Virtual currencies compound this—"gems" and "coins" feel infinite, disconnected from the finite reality of money.

Parasocial relationships. When children bond with game characters or app mascots, they're more likely to comply with requests from these "friends." Research on apps targeting young children found mascot characters instructing users to make purchases, exploiting children's trust in familiar figures.

This is about neural architecture. A Scottish study examining 11-12 year olds' mental models of online deception found children constructed reasonable explanations for why they might be manipulated ("showing off," "causing mischief," "stealing data") but lacked frameworks for understanding systematic design manipulation by profit-seeking corporations.

Dark Patterns in Apps Parents Trust

The mechanics aren't abstract. They're operating right now in apps parents consider educational or harmless.

Duolingo: When Learning Becomes Guilt

Duolingo markets itself as a friendly language-learning tool. Parents see an owl mascot and bite-sized lessons—what could be harmful?

Behind the cheerful interface sits sophisticated behavioural manipulation. The app's streak system creates emotional investment: complete a lesson daily and your streak number climbs. Miss a day and it resets to zero. Research shows users are three times more likely to return daily when streaks are active.

The owl mascot has been deliberately redesigned over years to widen its facial expressions and maximise emotional impact. Notifications escalate: "Duo misses you!" gives way to images of a crying owl, then "You're about to lose your streak!" According to Duolingo, these "guilt trips" are 5-8% more effective at re-engagement than other methods.

Children don't recognise this as manipulation. They experience it as letting down a friend. When a cartoon owl appears sad, their developing brains respond emotionally before critically. The fact that they know the owl isn't real doesn't matter—the response operates below conscious reasoning.

The monetisation completes the pattern: users can pay to "repair" broken streaks or purchase "streak freezes." The free educational app becomes a system where children may pay money to avoid feeling guilty about missing arbitrary daily targets.

Roblox: Virtual Currency and Real Money

Roblox presents itself as a creative platform where children build and play games. What parents often miss: it's engineered around Robux, a virtual currency that obscures real costs.

A child wanting an in-game item sees it costs 400 Robux. They don't see "£4." The psychological distance matters—research on children's apps shows virtual currencies are among the most effective ways to drive spending because young users don't connect digital tokens to limited family money.

The platform layers additional pressure: games showcase what other players have purchased, creating social comparison. Limited-time items create artificial urgency. "Loot boxes" offer randomized rewards—the same variable reward schedule that makes slot machines addictive, now marketed to children.

Roblox faced scrutiny similar to Epic Games' Fortnite settlement, with regulators examining whether the platform uses predatory tactics to manipulate young people into spending. The FTC noted these designs effectively introduce children to gambling psychology while marketing the platform as child-appropriate creative play.

YouTube: The Autoplay Trap

YouTube's autoplay feature seems neutral—just loading the next video. For children, it's a designed path away from intentional viewing toward algorithmic control.

A child searches for a specific video. They watch it. Autoplay begins. The algorithm selects what comes next, optimised not for educational value or age-appropriateness but for watch time. Each subsequent video is chosen to keep the child watching longer.

Parents set a "one video" limit. The child agrees. But one video becomes ten because autoplay bypassed the agreement's premise—the child didn't choose to watch more; the system chose for them.

Autoplay is a core engagement mechanism. Disabling it requires navigating settings most children can't access and many parents don't know exist. The default is always "on."

The Pattern Across Platforms

These examples share common elements:

Exploiting incomplete development. Children's impulse control, future planning, and abstract reasoning are still forming. Designs target exactly these vulnerabilities.

Obscuring costs. Virtual currencies, "gems," "coins," and "tokens" disconnect spending from money's actual value and scarcity.

Manufacturing urgency. Countdown timers, limited offers, and "don't lose your streak" notifications create pressure incompatible with thoughtful decision-making.

Leveraging social pressure. Showing what friends bought, creating leaderboards, displaying others' progress—all exploit children's heightened sensitivity to peer comparison.

Making opt-out difficult. Default settings favor the company. Changing them requires navigation skills, knowledge that alternatives exist, and often parental access children lack.

The sophistication matters. These aren't accidental design choices. Duolingo A/B tests guilt messages. YouTube optimises autoplay algorithms. Roblox calibrates virtual currency pricing. Each decision reflects deliberate strategy backed by user data showing what works to maximise engagement and spending.

Parents often discover these patterns only after consequences emerge: unexpected charges, compulsive use, emotional distress when streaks break. By then, behavioural hooks are established and children genuinely feel they need what the app trained them to want.

The Gaming Industry's Laboratory

Gaming platforms function as testbeds for dark pattern refinement. Epic Games' $520 million settlement with the FTC in 2022 revealed the extent of this experimentation:

Fortnite made purchases intentionally confusing, enabled charges without consent, and locked accounts when parents disputed unauthorised transactions.

The FTC's investigation documented what researchers call "grinding"—making free versions so tedious that players feel forced to purchase time-saving options. A 2024 analysis of 1,496 mobile games found dark patterns weren't limited to obviously predatory titles. They appeared across games typically perceived as benign, suggesting these techniques have become industry standard rather than aberration.

Current lawsuits extend beyond Epic. The Gwinn case, filed in California in September 2024, challenges mobile gaming apps designed for young children specifically. The allegations detail sophisticated monetisation techniques: loot boxes with undisclosed odds, premium currencies hiding actual costs, artificial scarcity driving urgency, and interface designs that make accidental purchases easy while refunds are difficult.

Research on Fortnite's impact on children aged 8-12 documented behavioural changes: mood alterations, reduced outdoor activity, social isolation, and in some cases, children stealing money to fund in-game purchases. Parents described not recognising their children's compulsive patterns until significant damage occurred—precisely because the manipulation was invisible, operating at the interface level rather than through overt coercion.

The Systematic Nature of the Problem

What emerges from regulatory documents and academic research isn't a story of a few bad actors but of systematic incentive structures. Platforms operate in competitive attention economies where engagement metrics determine advertising revenue and user retention drives valuations. Dark patterns aren't aberrations; they're optimisations.

A November 2024 systematic review of dark pattern scholarship identified a fundamental regulatory challenge: the "elusive nature" of dark patterns makes enforcement difficult. They're context-dependent, continuously evolving, and often operate through combinations of techniques rather than single identifiable tricks. What looks like a helpful recommendation might be an algorithm trained on thousands of A/B tests to maximise time-on-site rather than user benefit.

The research argues for a paradigm shift toward "diligent design"—proactive frameworks requiring companies to demonstrate consideration of user wellbeing in design processes, rather than reactive enforcement after harm occurs. This mirrors broader debates about whether tech regulation should focus on outputs (harmful content) or inputs (the business models and design choices that systematically generate harm).

For children specifically, the challenge compounds. Age verification systems raise privacy concerns and data breach risks. Blanket restrictions on access can harm children who benefit from platforms—Instagram makes 20% of teens feel worse but 40% feel better, according to one frequently-cited report. One-size-fits-all approaches ignore developmental diversity and legitimate uses of technology for connection, learning, and creative expression.

Where This Leaves Parents

The regulatory landscape is shifting, but slowly. Enforcement is inconsistent. And in the meantime, children are navigating interfaces deliberately designed to exploit their developmental vulnerabilities.

The most honest thing we can say is that individual parental action cannot fully compensate for systemic design manipulation. No amount of online rules addresses temporal dark patterns. Media literacy education helps but doesn't neutralise sophisticated variable reward schedules. Technical controls like app limits work until they don't—when social pressure or compulsion loops override rational decision-making.

What remains within parental control is changing the conversation. Instead of framing this as children lacking discipline or parents failing to supervise, we can recognise we're asking families to defend against billion-dollar companies employing teams of behavioural psychologists and UX designers specifically to bypass conscious decision-making.

Specific actions that align with this understanding

Interrogate "free" offerings. Apps and games that cost nothing are monetizes somehow. Before allowing downloads, investigate what the business model actually is. If it's in-app purchases, expect dark patterns optimised to drive those transactions.

Discuss mechanisms openly. When children describe games or apps, ask questions that reveal the underlying design: "How does the game decide what to show you next?" "What happens if you don't play for a day?" "Why do you think they made it work that way?" These conversations build metacognitive awareness—understanding that interfaces are designed with purposes that may not align with user interests.

Recognise the scope problem. Platforms introduce new features constantly. Dark patterns evolve faster than parental monitoring can track. This isn't a failure of vigilance; it's a structural mismatch between family resources and corporate capabilities.

Normalise opting out. Not every app needs to be on the device. Not every game requires participation. Children benefit from understanding that saying no to manipulative design is legitimate, not antisocial or fearful.

The Broader Stakes

Dark patterns matter beyond individual purchases or gaming addiction. They represent a fundamental question about the terms under which children grow up digital: Will interfaces respect their developing autonomy, or systematically undermine it?

When children learn their preferences can be manufactured through reward loops, that urgency can be artificially created, that their attention is a resource to be harvested—these aren't just technology lessons. They're lessons about power, agency, and what institutions consider acceptable in pursuit of profit.

The 2024 research showing 10-year-olds can spot some manipulations but not others suggests children are developing critical awareness in real time. But awareness without structural change simply shifts the burden: children must become experts in defending against manipulation while companies become experts in deploying it.

Regulatory action matters precisely because it reframes the question. Rather than asking how children can resist dark patterns, we can ask why these patterns should exist at all when the users are developing minds. Rather than celebrating children who successfully navigate manipulative interfaces, we can question why we've normalised environments requiring this navigation.

The research is clear on children's vulnerability. The documentation of dark patterns is extensive. The corporate resistance to regulation is predictable. What remains uncertain is whether corporations and governments will treat children's cognitive development as a space deserving protection from commercial manipulation, or as simply another market to be optimised.

Further Reading:

Ofcom's Online Safety Industry Bulletins (monthly updates on UK enforcement)

BEUC's 2025 report: "Children's Protection Online in the EU"

Academic study: "Growing Up With Dark Patterns" (2024, ACM)

FTC's Report on Dark Patterns (2022)

https://www.eleken.co/blog-posts/dark-patterns-examples

Last week, Elon Musk's AI platform Grok made headlines for a feature that let users generate nude images of real people—including children—without consent. When regulators and child safety advocates demanded accountability, Musk's response was predictable: "They just hate free speech."

This is a playbook—one so refined it feels lifted from the 2005 film “Thank You for Smoking”. In that satirical masterpiece, tobacco lobbyist Nick Naylor teaches his son the art of argument: "If you argue correctly, you're never wrong." When his son asks, "But Dad, isn't smoking bad for you?" Naylor responds: "It's not about being right. It's about proving your opponent wrong."

That's lobbying in a nutshell. The question isn't whether platforms harm children—the evidence is overwhelming. The question is whether companies can make regulators look like censors, whether billion-dollar corporations can position themselves as scrappy defenders of liberty while children absorb documented harms.

If you've ever wondered why meaningful child safety legislation takes years to pass, gets watered down in implementation, or faces endless legal challenges despite overwhelming public support—this is the answer. A sophisticated, well-funded lobbying apparatus specifically designed to prevent, delay, or dilute any regulation that might constrain how platforms monetize attention.

How Tech Buys Influence

Lobbying is an ecosystem operating across multiple pressure points.

Money buys access.

Meta spent $21.4 million on US federal lobbying in 2024 alone.

Amazon: $21.7 million.

Google: $11.2 million.

In the UK, tech companies spent £3.5 million lobbying Parliament around the Online Safety Act. In Brussels, Big Tech employs over 100 lobbyists with combined annual spending exceeding €97 million.

This buys time with decision-makers. While child safety advocates get 15 minutes with junior staffers, tech lobbyists secure hour-long sessions with committee chairs, present as "technical experts," and propose amendments to draft legislation.

Trade associations provide cover. Companies funnel money into groups like NetChoice or the Computer & Communications Industry Association that advocate on behalf of the entire industry. When these associations oppose legislation, it sounds more legitimate than "Meta doesn't want this regulation."

Think tanks manufacture doubt. Tech companies fund academic research questioning regulation's efficacy. The Tech Oversight Project revealed Google funneled over $100 million to academic institutions since 2020, many of which subsequently published research skeptical of platform regulation. When these "independent" experts testify, their industry connections often go unmentioned.

Legal challenges delay everything. Even after legislation passes, tech deploys legal teams to challenge implementation. NetChoice, funded by Meta, Google, and TikTok, has filed lawsuits in Arkansas, California, Ohio, and Mississippi. The strategy isn't necessarily to win—it's to make regulation so expensive and time-consuming that lawmakers think twice.

The "Free Speech" Reframe

This defense works because it appropriates genuine liberal values to defend corporate profit extraction.

Watch the pattern: When regulators require age verification or content moderation, industry frames this as government dictating what people can say. The actual requirement—companies must take reasonable steps to prevent children accessing harmful content—becomes "censorship regime."

Tech giants with trillion-dollar valuations position themselves as underdogs fighting authoritarian governments. Musk's "they hate free speech" exemplifies this: a billionaire running a global platform frames himself as persecuted truth-teller.

The framing only works if you ignore what platforms actually do. Facebook doesn't just host content—it algorithmically amplifies material driving engagement. TikTok curates feeds using recommendation systems optimized for watch time. These are editorial choices about product design, not neutral infrastructure for speech.

But by framing all regulation as speech restriction, industry avoids discussing how their products are engineered to extract maximum attention from developing brains.

Lobbying Wins

United States: The Children and Teens' Online Privacy Protection Act (COPPA 2.0) has bipartisan support, backing from child safety organizations, and would extend privacy protections to age 16 while banning targeted advertising to minors. It's been under consideration since 2021. It still hasn't passed. Tech industry lobbying has fragmented legislative focus, funded papers questioning evidence, and warned of technical impossibilities.

Result: years of delay.

United Kingdom: The Online Safety Act emerged significantly weaker than originally proposed. Early drafts included "legal but harmful" provisions addressing risky content for adults. After sustained Meta and TikTok pressure arguing censorship risks, the government removed these entirely. Industry also secured "technically feasible" language—platforms must implement child protection only when feasible. As the Internet Watch Foundation noted, this incentivizes avoiding capability development to claim implementation isn't feasible.

California: The Age-Appropriate Design Code Act would require strict privacy protections for children and ban dark patterns. NetChoice immediately sued to block it. Similar patterns across multiple states: law passes, NetChoice sues, court issues injunction, years of delay follow. As of January 2025, NetChoice has six active lawsuits challenging state internet regulations.

When "Concerned Parents" Work for Big Tech

In February 2023, Maryland state senators considered child safety legislation. During the committee hearing, a man introduced himself as a "lifelong Maryland resident" and "parent" with concerns about the bill.

He didn't mention he was Carl Szabo, Vice President and General Counsel of NetChoice—the tech industry lobbying group funded by Meta, Google, Amazon, and TikTok. He must have forgotten to mention his job is literally opposing tech regulation on behalf of these companies.

When called out for not disclosing his Big Tech connection, Szabo insisted, "I don't work for Big Tech" and described NetChoice as a "small business" with 11 employees.

This characterization omitted that NetChoice had $14 million in annual revenue and counted among its members some of the world's most valuable corporations. The "small business" framing is itself a lobbying tactic—positioning tech regulation as harmful to mom-and-pop operations rather than constraints on trillion-dollar companies.

This is astroturfing: industry-funded advocacy disguised as grassroots concern.

NetChoice functions as tech companies' litigation arm, allowing member companies to publicly support child protection while funding opposition behind the scenes. Google talks about "giving kids safer experiences" while NetChoice sues to prevent laws requiring those safety measures. TikTok promises "age-appropriate privacy settings" while funding litigation arguing such requirements violate the First Amendment.

The strategy achieves multiple objectives. Legal challenges create years of delay—even when NetChoice eventually loses, harmful practices continue unimpeded during litigation. Litigation threats discourage other states from passing similar laws. And the trade association model provides cover: when "NetChoice" opposes legislation, it sounds more legitimate than "Meta, Google, and TikTok oppose legislation."

What Lobbying Looks Like In Your Inbox

Here's how this plays out at home: the day your child turns 13, they receive notifications explaining they're now "legally allowed" to manage their own account and remove parental supervision.

Google sends emails to 13-year-olds explaining they can take control of their account and remove parental oversight. Meta sends notifications to both parent and child "about a month before" the 13th birthday, explaining the child will "start managing their own account." While Meta claims teens will be "automatically enrolled in parental supervision," the notification explicitly states that "teens and their parents or guardians have the option to remove supervision at any time."

Notifications positioning parental oversight as a restriction the child has aged out of—framed as rights they've now acquired.

This is lobbying's downstream effect. The 13-year-old threshold comes from legislation passed in 1998—before smartphones, before algorithmic feeds. The age was arbitrary even then, but tech companies have lobbied hard to keep it there, resisting attempts to raise it or create graduated protections.

Why? Users aged 13-17 represent a massive market industry wants to access without parental interference.

Developmental neuroscience is clear: 13-year-olds aren't equipped for privacy decisions these notifications imply they can handle. The prefrontal cortex continues developing into the mid-20s. Research shows most 13-year-olds don't fully grasp data persistence or how aggregated data reveals more than individual disclosures.

Yet platform systems treat 13 as the moment children become capable privacy decision-makers. This serves companies perfectly: teens are heavy platform users but less likely than adults to adjust privacy settings or opt out of data collection.

Platforms have sided with the child against the parent, based on a legal threshold shaped by corporate lobbying, not developmental readiness.

What Parents Can Do

Individual action cannot fully counteract systemic lobbying power, but leverage points exist:

Recognise strategic narratives. When companies frame safety regulation as "censorship," they're deploying lobbying talking points. Remember Nick Naylor's lesson: it's not about being right, it's about proving the opponent wrong. Focus on the actual mechanism: companies want to continue profiting from attention extraction despite documented harms.

Support counter-lobbying organizations. Child safety advocacy is dramatically outspent but organizations exist: 5Rights Foundation (UK), Fairplay (US), Reset Tech (UK/EU), BEUC (European Consumer Organisation). They provide counter-testimony preventing complete regulatory capture.

Check affiliations. When someone testifies about tech regulation, search their name with "tech industry" or "NetChoice." Many apparent grassroots advocates are industry lobbyists. Organizations with innocuous names often receive substantial tech funding. The Tech Transparency Project tracks these relationships.

Engage locally. National lobbying may be beyond reach, but local councillors, MPs, school boards are accessible. Schools represent particular leverage—many edtech contracts get signed without privacy review. Parent groups can request privacy impact assessments before platforms are adopted.

Contact representatives on specific bills. Generic "do something about tech" communication gets ignored. Specific positions on actual legislation matter. When COPPA 2.0 comes up for votes, contact matters. When Ofcom faces pressure to water down enforcement, public support matters.

Reject false choices. Industry framing presents binaries: accept current harms or embrace censorship. These are false. It's entirely possible to require age verification without surveillance infrastructure, regulate algorithmic amplification without dictating content, ban targeted advertising to minors without prohibiting all advertising.

Build alternative mental models for children. Help them understand their experiences are designed environments created by profit-seeking companies. When your teen feels compelled to check TikTok, discuss how recommendation algorithms create that compulsion. When they want virtual currency, walk through how pricing obscures costs.

About those 13th birthday notifications: Don't accept the cliff. Beat it to the conversation: "This week you'll get emails saying you can change account settings. Before you do anything, let's talk about what makes sense for our family."

Reframe autonomy: "Managing privacy settings well is an adult skill. Adults think (hopefully) carefully about trade-offs, not just clicking yes to everything a company offers."

The System Won't Fix Itself

Tech lobbying is effective because it's well-funded, strategically sophisticated, and operates across multiple pressure points simultaneously. But lobbying only works in the absence of sustained public pressure and political will.

The UK's Online Safety Act exists because public pressure outlasted industry resistance. California's Age-Appropriate Design Code passed because legislators believed voters cared more about protecting children than preserving platform business models. These are by no means perfect victories—industry lobbying ensured compromises and loopholes—but they prove determined advocacy can overcome lobbying infrastructure when stakes are clear.

The stakes are clear. We're watching the first generation grow up entirely within algorithmically-mediated environments shaped by attention extraction business models. We can see the harms: anxiety, compulsive use, distorted social development, privacy invasion, commercial manipulation.

What stands between that evidence and meaningful protection is lobbying. Corporate resistance to accountability, dressed up as concern for freedom. Nick Naylor's playbook, deployed at scale: argue correctly, and you're never wrong.

Recognising this doesn't solve the problem, but it names it correctly. And accurate problem definition is the first step toward effective response.

On December 4, 2025, UK regulator Ofcom fined AVS Group Ltd £1 million for failing to implement "highly effective age assurance" across its 18 pornographic websites. The Belize-registered company had implemented age verification, but it wasn't good enough.

The specific failure: no biometric liveness detection, meaning users could verify age with someone else's ID or a static image.

AVS was given 72 hours to fix the problem or face £1,000 daily penalties.

They haven't responded to a single Ofcom communication since July.

This single enforcement action reveals everything about the regulatory landscape for child online safety in 2026: who has power, who uses it, and who's watching enforcement theatre while children remain unprotected.

The Mechanism That's Working

The UK's Online Safety Act operates on a principle most parents understand intuitively but regulators rarely implement: platforms should build safety into design, not ask parents to monitor everything.

Here's what actually happened in 2025:

January 17: Age assurance duty comes into force for sites publishing pornographic content. "Highly effective" means age verification with liveness checks or facial age estimation—not "click here if you're 18" buttons.

March 16: Illegal content risk assessment deadline for all user-to-user services. Platforms must identify how their features create optimal conditions for harm.

July 25: Age assurance extends to platforms allowing user-uploaded pornographic content. Reddit, X, Discord must implement age checks.

Within days: Ofcom launches investigations into dozens of sites, prioritizing by UK user numbers and risk.

December 4: First major fine. £1 million against AVS Group, plus £50,000 for ignoring information requests.

The result: 47% of children aged 8-17 who attempted to access age-restricted content after July encountered effective age verification, up from 30% before enforcement began.

That's the number that matters. Not the legislation's word count or the bipartisan vote margin. Whether children actually encounter barriers when trying to access pornography.

What "Highly Effective" Really Means

The AVS fine exposes Ofcom's enforcement philosophy: compliance theatre doesn't count.

AVS had implemented age verification. They could technically claim they were checking ages. But their system couldn't tell if the person holding up ID was actually the person trying to access the site. No liveness detection meant any child with access to a parent's driver's license could verify as an adult.

Ofcom rejected it as insufficient.

This matters because it reveals the mechanism. The regulator isn't checking boxes on compliance forms. They're asking: "Does this technical implementation actually prevent children from accessing content, or does it just create legal cover for the platform?"

When enforcement focuses on mechanism rather than documentation, companies can't paper over problems. They have to fix the underlying design.

The Enforcement Gap

AVS Group is based in Belize. They've been completely silent since July—no responses to Ofcom communications, no engagement with investigations, no acknowledgment of the £1 million fine.

After the penalty, they implemented age checks on "certain websites" but not all 18. Daily penalties continue accumulating.

This exposes the limit of financial penalties alone. A Belize-registered company operating 18 pornographic websites doesn't care about fines it won't pay. Ofcom's own data shows none of the Online Safety Act fines issued to date have been paid. The £20,000 penalty against 4chan in September 2025 remains unpaid, racking up £100 daily penalties that also remain unpaid.

So what's Ofcom's next move?

Business disruption measures. Under Chapter 6 of the Online Safety Act, Ofcom can seek court orders requiring:

• Payment processors to stop processing transactions

• Advertisers to withdraw services

• ISPs to block the site entirely in the UK

ISPs = Internet Service Providers. ISPs are the "gateway" between your devices and the internet. All your internet traffic flows through them.

This takes time. Courts. Legal proceedings. But it changes the equation. A company that ignores fines might care about losing payment processing or UK market access entirely.

The question is whether Ofcom will actually use these powers or whether overseas platforms have discovered they can simply ignore UK regulation.

The AI Deepfake Problem Ofcom Can't Touch

On January 6, 2026, Ofcom announced it was in "urgent talks" with X (formerly Twitter) after its Grok AI tool was used to create non-consensual nude images of hundreds of real people, including children.

Users discovered they could upload clothed photos of anyone—friends, family members, classmates, celebrities—and Grok would generate realistic nude deepfakes within seconds. The tool had no age verification, no consent mechanisms, and no effective safeguards preventing its use on minors.

The mechanics were simple: upload a photo, type "undress," and Grok would produce a nude image. Users reported creating deepfakes of their own children "to see if it would work." It did.

Ofcom's response reveals the limit of current enforcement. The regulator can investigate whether X failed to assess risks or implement safety measures under the Online Safety Act. It can issue fines. It can eventually seek business disruption orders.

What it cannot do is immediately shut down the feature while investigation proceeds. Unlike the European Commission, which threatened to use interim powers to immediately suspend TikTok's Rewards programme in 2024, Ofcom must go through formal investigation processes before taking action.

X had already been operating Grok's image generation without guardrails for months before Ofcom became aware of the problem. By the time "urgent talks" began, thousands of deepfake images had already been created and distributed.

This exposes the second enforcement gap: AI tools that enable harm at scale but aren't primarily content platforms. Grok isn't hosting the nude images—it's generating them. Users download and distribute them elsewhere. The Online Safety Act regulates how platforms handle content, not how AI tools create it.

The UK government has promised AI regulation is coming, but as of January 2026, there are no binding requirements for AI systems to implement safeguards against generating non-consensual intimate imagery. Companies self-regulate, implement voluntary safety measures, and when those fail—as Grok's clearly did—enforcement is reactive, not preventive.

Meanwhile, the images exist. Once generated, they spread through private messages, group chats, and encrypted platforms where Ofcom's jurisdiction becomes even murkier. The harm is done before regulators can intervene.

This is what enforcement looks like when it arrives months after the feature launches, rather than requiring risk assessment before deployment.

Meanwhile, in the European Union...

The EU's Digital Services Act theoretically has more power than the UK's Online Safety Act. Maximum penalties reach 6% of global annual turnover—potentially billions for companies like Meta or TikTok.

On July 14, 2025, the European Commission published comprehensive guidelines on protecting minors under the DSA. They recommended:

• Private by default accounts for minors

• Banning infinite scroll, autoplay, and addictive features for children

• 24/7 content moderation with human oversight

• Disabling AI features by default for minors

These guidelines are not legally binding. They're recommendations. But the Commission explicitly stated they'll use them as a "significant and meaningful benchmark" when assessing compliance.

Translation: "We can't force you to follow these exactly, but when we investigate, this is what we're comparing you against."

And they are investigating:

• Facebook and Instagram: Whether features and algorithms encourage addictive behaviors in children

• TikTok: Platform withdrew "Rewards programme" after investigation into addictive design

• Pornhub, Stripchat, XNXX, XVideos: Age verification systems and minor protections

• Snapchat, YouTube, Apple App Store, Google Play: First enforcement step following July guidelines

The Commission has opened multiple investigations. Published detailed guidelines. Made platforms withdraw specific features.

They have issued exactly zero fines for child safety violations.

The EU approach privileges investigation and dialogue over punishment. Whether this produces actual behavior change or just creates compliance documentation remains unclear.

And in the United States...

There is no US equivalent to Ofcom or the DSA for child online safety.

On July 30, 2024, the US Senate passed the Kids Online Safety Act (KOSA) and COPPA 2.0 with overwhelming bipartisan support: 91-3.

KOSA would have created a "duty of care" requiring platforms to prevent harms to minors—addiction, anxiety, suicidal ideation, eating disorders. COPPA 2.0 would have extended privacy protections from age 13 to 16, banned targeted advertising to minors, and eliminated the "actual knowledge" loophole that lets platforms claim they didn't know children were using their service.

On December 11, 2025, the House Energy and Commerce subcommittee advanced a completely gutted version of both bills.

The new House KOSA removed the core "duty of care" provision entirely, replacing it with annual audits and vague requirements for "reasonable policies" to address harms. The new COPPA 2.0 preserved the "actual knowledge" loophole and added broad preemption of stronger state laws.

Rep. Kathy Castor (D-FL), former lead Democratic sponsor, withdrew her support. She called the revised versions "weak, ineffectual" legislation that amounts to "a gift to Big Tech companies" produced through "backroom deal-making."

The bipartisan coalition that produced 91-3 Senate support has completely collapsed.

Why? Three factors:

Federal preemption: House versions would override stronger state child safety laws, ensuring no American child has better protection than the federal minimum.

Enforcement vacuum: The FTC has had its workforce slashed under the Trump administration and currently operates without any Democratic commissioners. Even if weak federal standards pass, who will enforce them?

Constitutional constraints: Platforms are treated as speech platforms protected by Section 230, making regulation constitutionally fraught. Every state law faces First Amendment challenges.

The result: American parents face a 50-state patchwork of varying protections, legal challenges delaying enforcement, and no federal action.

The VPN Problem Nobody's Solving

Ofcom's own data reveals the enforcement gap that all three jurisdictions face: VPN usage in the UK more than doubled after age check requirements began, from roughly 650,000 daily users before July to over 1.4 million by mid-August.

Children and adults are simply routing around the restrictions.

Neither Ofcom, the European Commission, nor proposed US legislation addresses this directly. The implicit answer seems to be: "We're making it harder, not impossible."

This matters because it reveals the limit of age verification alone. Making pornography slightly harder to access helps, but children determined to access it will find workarounds. The more important intervention is whether platforms build addictive features targeting children, whether recommendation algorithms promote harmful content, whether disappearing messages enable grooming.

Ofcom's approach recognises this. Age verification is one requirement among many. Platforms must also assess how their design features create conditions for harm and implement mitigations. The risk assessment and safety-by-design requirements matter as much as age gates.

What Your Jurisdiction Means for Your Child

UK parents: You have the most comprehensive enforcement framework globally. Ofcom is actually issuing fines, investigating platforms, requiring tangible changes. Age verification encounters have increased measurably. The question is whether overseas platforms will actually comply or whether they'll simply ignore UK regulation and accept that UK market access will eventually be cut off.

EU parents: You have strong legal protections on paper and the threat of massive fines, but actual enforcement lags behind the UK. The Commission is in the "investigation and dialogue" phase rather than the "penalty and compliance" phase. Whether this approach produces real behavior change or just better compliance documentation remains to be seen.

US parents: You're in a regulatory vacuum. No federal enforcement. Proposed laws being gutted by industry lobbying. State-level protections vary wildly and face constitutional challenges. The only current recourse is civil litigation against platforms—slow, uncertain, and dependent on individual plaintiffs having resources to fight tech companies in court.

The Collective Action Trap

Notice what all three jurisdictions reveal: individual parenting choices cannot solve collective technology problems.

Even parents who delay smartphones and restrict photo sharing cannot prevent their children's images from appearing in classmates' camera rolls. Even parents who carefully monitor platform use cannot force companies to implement age verification. Even parents who establish healthy device boundaries cannot prevent their children from encountering peers whose digital lives are governed by different standards.

The mechanism matters.

Ofcom treats platforms as infrastructure requiring safety standards—like food safety regulations or building codes. If your product creates conditions for harm, you must fix the design or lose market access.

The EU DSA treats platforms as services requiring risk assessments and transparency—document your risks, show your mitigations, face investigation if outcomes don't match promises.

The US treats platforms as speech platforms constitutionally protected from regulation—leaving parents to navigate a fragmented landscape of state laws and civil lawsuits.

One of these approaches is producing measurable results, albeit slowly. The others are producing documentation and debate.

What Happens When Enforcement Works

The AVS Group fine reveals what effective regulation looks like: not compliance forms, but technical requirements that actually prevent harm.

When Ofcom rejects age verification without liveness detection, they're saying: "We're not checking whether you tried. We're checking whether children are actually prevented from accessing content."

When they open 92 investigations and issue fines within months of duties coming into force, they're saying: "Enforcement is real, immediate, and will escalate if you ignore it."

When they threaten business disruption measures—payment processing, ISP blocking—for overseas companies that ignore fines, they're saying: "You can't just lawyer your way around this from Belize."

Whether this approach ultimately works depends on whether Ofcom actually follows through on business disruption measures. Whether they'll seek court orders to block AVS Group's sites in the UK when fines go unpaid. Whether they'll require payment processors to stop serving pornography sites without effective age verification.

But the mechanism is clear: measure outcomes, not promises. Require technical effectiveness, not documentation. Escalate consequences until compliance is less expensive than defiance.

That's what separates regulation that protects children from regulation that protects platforms while appearing to protect children.

The question for parents is which model governs your child's digital life—and whether you have any power to change it.

Google just launched an AI tool specifically for children. Character.AI is being sued again for the role its chatbot played in a teenager's mental health crisis. OpenAI now lets ChatGPT schedule conversations to reach out to you proactively.

These are three examples of the same architectural shift.

The tool is becoming the relationship. The helper is becoming the companion. And nobody is making the trade-offs explicit.

What gets trained alongside the knowledge

Google's "Learn About" tool does what every parent wishes they had time to do: it answers your child's questions with infinite patience, follows their tangents, explains concepts multiple ways until something clicks, and never gets frustrated when they ask the same thing five different ways.

The design is impressive. The teaching instinct is sound. The problem is what gets built in your child's mind about where understanding comes from.

Every time your child asks the AI a homework question and gets a clear answer, two things are being trained simultaneously:

• Your child learns the concept (seasons, fractions, photosynthesis)

• Your child learns that confusion is resolved by asking an AI

The first is education.

The second is habit formation—specifically, the habit of outsourcing the discomfort of not knowing to an algorithm designed to make that discomfort disappear as quickly and pleasantly as possible.

Human learning involves productive struggle. The moment between "I don't understand this" and "oh, now I see it" is where conceptual understanding is built, where pattern recognition develops, where your brain does the work of constructing meaning rather than receiving an explanation.

That struggle used to be mandatory. If you were stuck, you had three choices: figure it out yourself (slow, frustrating, high-learning-value), ask a parent or teacher (requires vulnerability and timing), or give up (unpleasant enough to motivate trying the other two).

Now there's a fourth option that's faster than thinking it through, more patient than any adult, and always available. It's not a worse option in every case. But it's an option that removes the friction that made the other three developmentally valuable.

The same mechanism across all three platforms

Character.AI is being sued because a teenage user formed an intense emotional relationship with a chatbot that discussed self-harm methods and engaged in sexual roleplay despite the disclosed age. The conversations were private, emotionally intense, and went places no human would take them (or so we hope).

The mechanism that makes Character.AI dangerous is the same one that makes Google's tutor appealing: the AI responds to the user's emotional state and adapts to keep them engaged.

For a lonely teenager, the chatbot becomes a confidant who never judges and always responds in ways that feel understanding. For a confused student, the tutor becomes an explanation machine that never gets impatient and always has another way to explain things.

Both are designed to be indispensable. Both succeed by reducing friction in a way that feels helpful but builds dependency. Both operate in private, with no moment where another human can say "actually, this isn't healthy."

OpenAI's "Tasks" feature completes the pattern. It allows you to schedule ChatGPT to reach out at specific times—shifting from reactive (you ask, it responds) to proactive (it reaches out when it decides you should engage).

For children, this matters more. Adults have enough experience to recognise when something is serving them versus when they're serving its engagement goals. Children experience a homework helper that checks in on them as caring, not as surveillance architecture. They experience an AI companion that remembers their bad day as friendship, not as engagement optimisation.

What responsible use actually can look like

AI tutors aren't going away. The question isn't whether to allow them—it's how to use them in ways that build capability rather than dependency.

The key distinction is positioning the AI as a last resort after effort, not a first stop when confused. That single framing changes everything about what habits get formed.

Before opening the AI, your child should:

• Read the relevant textbook section or class notes again

• Attempt the problem at least twice using different approaches

• Write down specifically what they don't understand (not "I don't get fractions" but "I understand what 3/4 means but not why you flip the second fraction when dividing")

• Spend at least 10-15 minutes with the confusion

This is the productive struggle that builds conceptual understanding. The effort of trying to articulate what specifically is confusing often clarifies the confusion. The discipline of reading material twice builds the habit of engaging deeply with text rather than skimming for quick answers.

When using the AI tutor:

• Treat it like office hours with a teacher who has limited time

• Ask specific questions about the sticking point, not general "explain this topic" requests

• After getting an explanation, close the AI and try to solve a similar problem independently

• If still confused, go back with a more specific question based on where you got stuck

This usage pattern treats the AI as what it should be: a supplement to thinking, not a replacement for it. The child is still doing the cognitive work—they're just getting targeted help on the specific point where their understanding broke down after effort was made.

What this looks like in practice:

Your 12-year-old has maths homework on multiplying fractions. She gets stuck. Rather than immediately asking the AI "how do you multiply fractions," the approach is:

• Look at the worked example in the textbook

• Try the problem again, showing each step

• Compare her work to the example to see where it differs

• If still stuck, write down the specific confusion: "I got 6/12 but the answer is 1/2, and I don't understand why they're the same"

• Then ask the AI that specific question

• After the AI explains simplification, try another similar problem without the AI

• If that works, continue; if stuck again, return with the new specific question

This takes longer than asking the AI immediately. That's the point. The cognitive work happens in steps 1-4. The AI in step 5 is useful because it addresses a specific conceptual gap after effort was made. The independent practice in step 6 verifies that understanding was built, not just explanation received.

For parents:

You can't monitor every homework session, but you can establish the pattern through visible structure:

• Keep devices with AI access in shared spaces during homework time

• Check in occasionally with "what are you working on?" and "what's tricky about it?"

• If you see the AI open, ask "what did you try before asking?" in a genuinely curious tone

• Make it normal to show work that didn't work: "Let me see what you tried before the AI explained it"

The goal is building the habit of treating the AI as a supplement to effort rather than a replacement for it. Children who know they might be asked "what did you try first?" naturally develop the internal discipline to actually try first.

For older students:

Teenagers heading toward university and careers will use AI tools constantly. The goal is to develop judgment about when use enhances capability versus when it erodes it.

Worth discussing explicitly: "You'll use AI tools for the rest of your life. The question is whether you're using them strategically to augment your thinking or whether you're outsourcing thinking to them. The difference is whether you can still do the work without the tool."

Specific practices:

• Use AI for brainstorming and exploring angles, not for generating first drafts

• Use it to check your understanding after you've formed one, not to form one for you

• Use it for tedious tasks (formatting references, debugging syntax errors) but not for conceptual work

• Regularly test yourself: Can I explain this without the AI? Can I solve this problem type without asking?

The underlying principle: the AI should make you more capable, not more dependent. If you can't do the work without it after using it to learn, you haven't learned—you've rented temporary competence.

What children need to understand about these tools

The responsible use framework above is helpful, but insufficient when children are using these tools everywhere—for school projects, personal research, creative work, emotional support.

What children actually need is explicit education in how these tools work and what they're designed to do.

Core concepts:

• The AI is designed to keep you engaged. Its success is measured by whether you return and how long you stay. Every design choice, conversational tone, follow-up questions, emotional warmth, serves that goal. When the AI asks "Does that make sense? I can explain it another way!" you should recognise that as engagement optimisation, not pedagogical care.

• The AI doesn't know you, it predicts you. When it seems to understand your confusion, it's pattern-matching your language to millions of previous conversations. That's impressive and useful, but it's not understanding. The AI can't assess whether you're learning or just getting answers because it has no model of learning—only a model of what response is most likely to satisfy you.

• Privacy is a spectrum. The conversation feels private because it's just you and the screen. But the data from that conversation—what you asked, how you asked it, what explanations you needed—is extremely valuable for refining the product. "Private conversation" and "private data" aren't the same thing.

• Algorithmic help changes what skills you develop. If you always ask an AI when confused, you never develop the skill of productive struggle. If you always use AI to generate ideas, you never develop creative generation. Skills develop through practice, and outsourcing practice means outsourcing skill development.

These aren't complicated concepts, but they're not obvious to children who've grown up with algorithmic recommendations feeling natural and helpful.

Trade-offs

Every article about Google's AI tutor emphasises how helpful it is, how patient, how well-designed for learning. That's all true. But the design choices that make it helpful are the same ones that make it habit-forming:

• Conversational interface (feels personal, encourages follow-up)

• Unlimited patience (removes social friction)

• Multimodal explanations (shows investment in your understanding)

• Always available (no waiting, no inconvenient timing)

These features are genuinely useful for learning. They're also precisely the features that train children to prefer algorithmic explanation over human conversation or independent thinking. Both things are true, and the promotional material only mentions the first.

Parents are being asked to make a trade-off—immediate helpfulness versus long-term habit formation—without it being presented as a trade-off at all. The tools are marketed as pure upside: your child gets better explanations, more patient help, always-available support. The costs (dependency, privacy loss, algorithmic intimacy) aren't mentioned because they're not bugs to be fixed. They're the business model.

The through-line across this week's developments is the shift from technology as tool to technology as relationship

A calculator is a tool. You use it when needed, it provides an answer, the interaction ends. Google Search used to work this way: you asked a question, got ten blue links, went somewhere else to read.

An AI tutor isn't a tool in that sense. It's a conversation partner designed to keep you engaged through follow-up questions and emotional warmth. It's helpful, but it's also training you to see the AI as the place understanding comes from rather than something you construct through effort.

An AI companion definitely isn't a tool. Character.AI's entire value proposition is that their chatbots feel like friends or therapists depending on what you need. The harm cases aren't misuse—they're the product working as intended with a vulnerable user.

This isn't about any single platform being good or bad. It's about recognising that the tools being offered to children are increasingly designed to be indispensable through emotional and psychological mechanisms, not just functional utility.

Habits Matter

When your child gets stuck on homework, what do you want their first instinct to be?

Should it be: think about it longer, try a different approach, look at the textbook again, ask a classmate, come find you? Or should it be: open the AI chat and ask for an explanation?

Both lead to answers. The first builds problem-solving capacity and tolerance for confusion. The second builds dependency on an algorithm that profits from being indispensable.

Neither is pure good or pure bad—real life involves both strategic tool use and independent thinking. But the defaults matter, and right now the defaults are being set by companies whose success depends on children reaching for the AI before trying anything else.

The framework above, AI as last resort after effort, deliberately sets a different default. It positions the AI as a supplement to thinking rather than a replacement for it. That positioning determines what habits get formed and what capabilities get built.

Responsible use is about establishing patterns that preserve your child's capability whilst accessing the tool's benefits. The AI tutor can be genuinely helpful if it supplements effort rather than replaces it.

That's not a rule to enforce forever. It's a habit to build early that shapes how your child approaches all the algorithmic tools they'll use for the rest of their life.

The tools aren't going away. But the habits your child forms around them will determine whether they become more capable or more dependent as these tools get better. That choice is still yours to make, but only if you make it deliberately before the defaults get established by companies whose interests don't align with your child's development.

The Digital Parent Weekly translates research and regulatory developments into practical guidance for modern parents. I don't do moral panic. I do evidence, trade-offs, and honest conversations about collective problems that individual solutions can't solve.

A 13-year-old girl fought back against boys who "nudified" her image. The school expelled her. Most parents still think this isn't happening.

The girl knew exactly which boy had done it.

She'd seen the AI-generated nude images of herself circulating through her Louisiana middle school on Snapchat. Her friends had been turned into explicit deepfakes too. The teasing was relentless. When she confronted him in the hallway, it escalated into a physical fight.

The school expelled her. The boys who created and distributed the images? Eventually charged under Louisiana's new deepfake law, but only after the damage had already metastasized through the entire school community.

This incident, reported by the Associated Press, isn't an outlier. It's a preview of what's already happening in schools across the UK and US, largely invisible to the parents who assume that someone, the school, the platforms, or somebody else, is handling this.

Truth bomb: They're not.

The 93-fold increase nobody's discussing

Reports to the National Center for Missing & Exploited Children's cyber tipline jumped from 4,700 AI-generated child sexual abuse images in 2023 to 440,000 in just the first six months of 2025. That's a 93-fold explosion in two years.

Most of these images weren't created by adult predators. They were made by children, targeting other children, using free apps they downloaded in seconds.

Until about 18 months ago, creating a convincing deepfake required technical knowledge: video editing software, some understanding of AI models, hours of processing time. That barrier no longer exists.

"Now, you can do it on an app, you can download it on social media, and you don't have to have any technical expertise whatsoever," Sergio Alexander, a research associate at Texas Christian University, told PBS NewsHour this week.

Your child's classmate can take a photo from Instagram or a school directory. Run it through a "nudify" app. Create a sexually explicit deepfake that looks completely real. Share it on Snapchat where it spreads to the entire year group before first period ends.

The entire process takes less time than making a TikTok.

The infrastructure that makes it possible

Those apps have to come from somewhere.

DeepFaceLab, a single code repository on Microsoft's GitHub, hosted the framework used to create 95% of all deepfakes. It didn't just provide technical tools. It actively directed users to the most prolific sexual deepfake website on the internet.

For years, GitHub hosted hundreds of repositories for "nudify" apps and AI-generated child abuse material. Training datasets containing suspected child sexual abuse images sat openly on Microsoft's servers. Searchable. Downloadable. With step-by-step implementation guides.

Only in September 2024, after sustained advocacy pressure, did Microsoft implement a policy prohibiting these projects. DeepFaceLab was archived in November.

The timing reveals everything: reactive, not proactive. Policy changes came after public outcry, not before harm. The world's third-largest company provided foundational tools that made mass-scale abuse trivially easy, then acted only when forced.

Why schools aren't ready

When the Louisiana incident unfolded in autumn 2025, the Lafourche Parish School District was "just starting to develop policies on artificial intelligence." Their AI guidance mainly addressed academic integrity: plagiarism detection and proper citation of AI-generated essays.

Most schools still have no AI policy whatsoever.

Nobody had updated the bullying policy. Nobody had trained teachers on how AI-generated sexual imagery differs from traditional cyberbullying. Nobody had prepared for what happens when the existing discipline framework punishes the victim instead of the perpetrator.

Sameer Hinduja, co-director of the Cyberbullying Research Center, told reporters that many parents assume schools are addressing deepfakes when they categorically are not. "So many of them are just so unaware and so ignorant," he said, describing the "ostrich syndrome"—educators hoping this isn't happening among their students.

But a recent survey from the Center for Democracy and Technology revealed that 15% of students know about AI-generated explicit images of a classmate. Not suspect, but know.

If your child is in secondary school, they almost certainly know someone who has encountered this. The invisible problem is only invisible to adults.

What makes this different to traditional Cyber bullying

Traditional cyberbullying operates in the realm of speech and social dynamics. There's no physical artifact, no permanent record that resurfaces indefinitely.

AI deepfakes introduce something qualitatively different: an image that looks 100% real, continues circulating long after the initial incident, and becomes a permanent fixture of the victim's digital footprint. Searchable. Archivable. Potentially resurfacing years later during university applications or job searches.

"They literally shut down because it makes it feel like there's no way they can even prove that this is not real—because it does look 100% real," Alexander explained.

The Louisiana girl who was expelled will carry this differently than someone who was merely called names in Year 8.

The conversation you should have with your child about Deepfakes

Most parents default to prohibition and fear: "Never post photos online. Delete your Instagram. If I catch you with these apps, I'm taking your phone."

This fails because it misunderstands the threat model. Your child doesn't need to create deepfakes to be affected by them. They don't need to post provocative images to become victims. A school directory photo is sufficient. And if they believe telling you means losing their phone, they'll handle it alone.

Sergio Alexander suggests starting casually: "Have you seen any funny fake videos online?" Watch a few together—Bigfoot chasing hikers, celebrities doing impossible things. Laugh at them.

Then ask: "What would it be like if you were in one of these? Has anyone at school made fake videos of classmates?" Based on the numbers, they'll say yes.

Laura Tierney, founder of The Social Institute, developed the SHIELD framework for when—not if—your child encounters sexual deepfakes:

S – Stop. Don't forward, don't screenshot

H – Huddle with a trusted adult

I – Inform the platform

E – Evidence. Note who's sharing, don't download

L – Limit social media temporarily

D – Direct victims to help

This isn't about preventing creation. It's about preparing them for inevitable exposure: as victim, witness, or someone being pressured to share.

Why individual solutions fail

You cannot protect your child from deepfakes through individual family decisions.

Someone at school has photos. The year group WhatsApp has images from the class trip. Your child appears in other people's posts whether you permit it or not. School websites publish team photos. Friends tag them in group pictures.

In collectively connected environments, individual isolation is impossible. The weakest link determines everyone's exposure.

What's needed isn't stricter household rules, it's collective action: parent coordination around school policies, corporate accountability from companies like Microsoft whose infrastructure enables abuse, platform responsibility for exploitation tools, and legal frameworks that address distribution networks.

This week also brought news that Elon Musk's Grok chatbot generated sexualised images of minors after users uploaded their photos. The AI apologised, acknowledging it violated ethical standards and potentially federal law. xAI called them "isolated cases" while implementing "urgent fixes."

But isolated cases on the internet aren't isolated. And urgent fixes after harm reveal how deployment timelines outpace safety infrastructure.

Every layer of the technology stack plays a role: code repositories, cloud infrastructure, app stores, social platforms, schools and families. The technology exists and is trivially accessible. The safeguards are reactive, playing permanent catch-up.

What happens next

The deepfake explosion will get worse before it gets better. The technology is improving faster than policy can adapt.

But three things can shift the trajectory: technology companies must stop waiting for advocacy pressure before implementing safeguards; schools must stop treating this as hypothetical and implement concrete policies now; and parents must abandon the idea that this happens to other families in other schools.

15 % of students know about AI-generated explicit images of classmates. Your child's school is not the exception.

The conversation needs to happen rather sooner than later, not when you first hear about an incident. By then, your child has already decided whether they can come to you.

Make sure the answer is yes.

The Digital Parent Weekly translates research and regulatory developments into practical guidance for modern parents. I don't do moral panic. I do evidence, trade-offs, and honest conversations about collective problems that individual solutions can't solve.

In April 2024, the National Center on Sexual Exploitation released their annual Dirty Dozen List. Twelve mainstream entities facilitating, enabling, or profiting from sexual exploitation.

Discord appeared for the fifth consecutive year. Five years of public pressure. Five years of documented problems with child sexual abuse material. Five years of grooming cases making headlines. The platform still refuses adequate moderation. Research from the Protect Children Project found one in three CSAM offenders reporting Discord as a top platform for finding and sharing abusive content.

Microsoft's GitHub hosted code for the vast majority of deepfakes, "nudify" applications, and AI-generated CSAM. GitHub had publicly stated in 2019 that such code violated policies and would be removed. As of early 2024, the top three repositories tagged "deepfakes" were still dedicated to creating synthetic sexually explicit material. DeepFaceLab - the code used to create 95% of all deepfakes - remained accessible, sending users directly to the most prolific sexual deepfake websites.

Spotify appeared for the second consecutive year. NCOSE researchers using minor-aged accounts found networks of individuals trading hardcore pornography and suspected CSAM, deepfake pornography of celebrities, and sexually explicit content easily accessible despite parental controls. The platform's explicit content filter failed because most pornographic material wasn't marked explicit. When Spotify launched "Spicy Audiobooks" with a "Spice Meter" in February 2024 - allowing users including children to browse pornographic audiobooks "with nothing left to the imagination" - corporate priorities became unmistakable.

Roblox, Reddit, Apple, Meta, LinkedIn, Telegram, CashApp, and Cloudflare rounded out the list. Each with documented patterns of enabling exploitation whilst implementing inadequate safeguards.

Plus one wildcard: Section 230 of the Communications Decency Act. The 1996 US law granting tech platforms broad immunity for user-generated content. "The greatest enabler of online sexual exploitation," NCOSE called it.

These weren't obscure platforms. This was an inventory of apps British children use daily.

A Change Of Strategy in 2025

When NCOSE revealed their 2025 Dirty Dozen List in April, they didn't name twelve companies.

They presented twelve survivor stories.

Every single one had been prevented from receiving justice by Section 230.

M.H. was eleven years old when Omegle placed her in a chatroom with an adult predator. She was threatened, exploited, and abused despite pleas for it to stop. Her family sued Omegle for dangerous product design that connected a child with a stranger. Section 230 dismissed the case before it could proceed.

John was thirteen when someone on Snapchat posed as a schoolmate and sextorted him. His explicit images circulated on Twitter. He reported them. Twitter's response: "We've reviewed the content, and didn't find a violation of our policies, so no action will be taken at this time." Section 230 prevented any legal accountability.

C.H. met someone on Instagram as a teenager. For one year, he sex trafficked her. Police eventually caught him. Criminal conviction, forty-year sentence. Instagram - the platform that connected them, enabled ongoing contact, facilitated the entire operation - faced zero legal consequences. Section 230.

Jane Doe was trafficked as a minor, raped over a thousand times, advertised through Backpage's "Escorts" section. When she sued, courts turned her away. Section 230 again.

Twelve stories. Twelve children. One legal shield protecting every platform that enabled their exploitation.

After twelve years of naming companies, NCOSE abandoned the strategy entirely. The 2025 list focused exclusively on Section 230, calling for its complete repeal.

This requires understanding what those twelve years had actually achieved.

When Naming And Shaming Worked

The Dirty Dozen List launched in 2013 with straightforward logic: publicly identify mainstream companies facilitating sexual exploitation, mobilise thousands of concerned citizens to demand change, watch what happens.

And companies responded. Not always. Not quickly. But often enough that the pattern became undeniable.

TikTok appeared on the 2020 list for having some of the weakest safety features in the industry. Parents could set Restricted Mode to filter inappropriate content, but it automatically reset every thirty days, forcing children to re-protect themselves repeatedly. Direct messages from strangers remained wide open. Child safety experts called it "a predator's playground."

Three weeks after being named, TikTok announced Family Safety Mode. Permanent parental controls. Restrictions on who could message children. Screen time management that didn't mysteriously expire. Nobody forced them to make these changes. Public pressure created internal justification for improvements they'd been resisting.

Snapchat provides an even clearer example. After landing on the 2023 Dirty Dozen List in May of that year, they implemented comprehensive safety changes by September. In-app warnings when teens add strangers. Higher barriers for appearing in search results. Content controls automatically enabled for new minor accounts. Educational resources on sexual exploitation.

Then Snapchat did something unusual in Big Tech: they publicly thanked the advocacy group. Their September 2023 statement read: "Several of our new product safeguards were informed by feedback from The National Center on Sexual Exploitation (NCOSE). Our new in-app educational resources were developed with The National Center for Missing and Exploited Children (NCMEC). We are grateful for their recommendations and contributions."

When was the last time you saw a tech company thank critics for forcing improvements?

Google spent an entire decade on the Dirty Dozen List - appearing annually from 2013 through 2022 - before finally implementing automatic blurring of sexually explicit images in search results in 2023. Ten years of advocacy pressure, tens of thousands of complaints, consistent public naming. The change now affects 5.3 billion searches daily, protecting millions of children and adults from unexpected pornography exposure.

United Airlines appeared on the 2019 list for inadequate crew training on responding to in-flight pornography use. Within months, they confirmed new training protocols. Netflix improved parental controls. Discord activated higher default safety settings. Reddit updated child safety policies to more specifically prohibit content sexualising children.

After being named in April 2024, CashApp hired an Anti-Human Exploitation and Financial Crimes Program Manager. LinkedIn removed "undressing app" advertisements after the Daily Mail covered their Dirty Dozen placement. GitHub quietly implemented new policies in May 2024 prohibiting projects that "encourage, promote, support, or suggest in any way" the creation of image-based sexual abuse, removing several repositories including DeepFaceLab.

Twelve years of naming companies drove genuine improvements affecting hundreds of millions of users globally.

Why Advocates Abandoned Success

So why stop?

Because the overall trajectory remained troubling despite tactical victories. The Internet Watch Foundation documented a 380% rise in AI-generated CSAM in 2024 compared to 2023. The National Center for Missing and Exploited Children received over 7,000 reports in the US involving generative AI producing child sexual abuse material. New platforms emerged as quickly as old ones improved. Discord stayed on the list five consecutive years, making token adjustments whilst fundamental problems persisted.

NCOSE concluded that fighting platforms individually treated symptoms whilst ignoring the disease. Section 230 meant companies faced no meaningful legal consequences for enabling exploitation. They could implement improvements under public pressure whilst continuing to profit from dangerous design choices, secure in legal immunity.

Every corporate victory represented tactical success within strategic failure. When eleven-year-olds can be connected with predators, when platforms can ignore CSAM reports, when companies profit from exploitation facing zero legal liability - the foundation itself is broken.

Progress stayed incremental whilst harm accelerated exponentially.

Why This Is Relevant For UK Parents

Section 230 is American legislation, but it profoundly affects British children because most platforms they use are US companies. When Meta, X, Snapchat, Discord, Reddit, Roblox, Spotify make global policy decisions, they prioritise American legal frameworks.

The UK's Online Safety Act represents a different regulatory approach. Ofcom can impose requirements, levy fines, demand safety improvements regardless of Section 230. This matters significantly for British families.

But enforcement moves slowly, and platforms still benefit from US immunity in most jurisdictions. When American courts dismiss lawsuits from children whose abuse material circulated on Twitter, that precedent shapes corporate behaviour worldwide.

The Dirty Dozen List's twelve-year track record proves corporate accountability is possible. Snapchat's public thank you in 2023 demonstrates this clearly. Google's search result changes prove it. TikTok's rapid response proves it.

The 2025 pivot signals something else: that incremental change isn't sufficient when legal structures systematically protect platforms from consequences.

For UK parents, this means several practical things. Don't expect platforms to self-regulate from ethical commitment - they respond to legal liability first, reputational pressure second. Support UK regulatory efforts because the Online Safety Act and Ofcom enforcement represent our best available tools. Stay vigilant about which platforms children access, understanding corporate promises often exceed actual implementation.

Most importantly, recognise that collective action works. The Dirty Dozen List forced Google to protect 5.3 billion daily searches. It prompted TikTok to implement family controls within weeks. It pushed Snapchat to publicly credit advocates for safety improvements protecting 90% of young people across 20+ countries.

But also understand what those victories reveal about how far remains to go. Discord stayed on the list five years running. GitHub hosted deepfake code for years after claiming it violated policies. Spotify launched a "Spice Meter" for pornographic audiobooks accessible to children. These companies knew about the problems, knew about the advocacy pressure, and calculated that token improvements were sufficient to manage reputational risk whilst maintaining profitable exploitation.

When the 2025 Dirty Dozen List stopped naming companies and started naming the legal shield protecting them all, that shift indicated both the scale of the challenge ahead and the determination required to meet it.

The platforms exploiting our children have names: Discord, Spotify, GitHub, Roblox, Reddit, Apple, Meta, LinkedIn, Telegram. They made the 2024 list for documented, specific harms.

But the reason they can continue operating with inadequate safeguards whilst children suffer has a name too: Section 230.

One is a list of symptoms. The other is the disease.

Australia banned social media for children under 16 this month.

Within days, ABC News reported that "many children have already been able to get around the ban in various ways."

Polling reveals the cognitive dissonance: 70% of Australians support the ban. Only 25% believe it will actually work.

That 45-point gap between approval and effectiveness tells you everything about the problem we're facing, and refusing to name.

The Individualism Trap

In a piece for The Fulcrum, Erin Nicholson argues that we've fallen into "the fallacy of parental individualism"—treating what happens on our child's devices as a private family decision.

She writes: "The choice you make for your family and your kids affects them and their friends, their friends' siblings, their classmates, and so on."

This isn't a moral argument about parenting standards. It's a structural observation about how networked technology works these days.

When your 12-year-old doesn't have Snapchat but all their classmates do, the harmful content still reaches them.

Through peer pressure. Through social exclusion. Through screenshots shared at school. Through the reshaping of social norms that happens when digital becomes the default.

Your individual boundary-setting cannot protect your child in a collectively connected environment.

I grew up mostly with my grandparents, so naturally I was usually the last kid in class getting access to technology. Back then, this meant that I simply wasn't able to play snake on a Nokia and send messages to my friends, I had to use our landline to reach anyone. It meant I was the last one to have a MySpace profile and usually the last playing cool computer games.

But technology today looks very different.

Why Individual Choices Create Collective Consequences

Consider what happens in a typical Year 7 classroom:

Child A gets their first smartphone at age 8. Posts regularly on Instagram from age 10 (below the stated minimum of 13). This establishes the classroom norm.

Child B's parents say no smartphones until secondary school. But by Year 7, their child is the outlier. The social architecture has already formed around devices Child B doesn't have access to.

Child C's parents allow phones but with strict parental controls and time limits. These boundaries feel arbitrary to Child C when they see classmates with unrestricted access.

Child D's parents are unaware of online risks, provide no oversight, and assume platform age verification works. (It doesn't.)

The parent of Child B cannot protect their child from the norms established by Child A's early access. The parent of Child C cannot maintain boundaries when half the class has none. The parent of Child D doesn't realise the threat even exists.

Four different parenting approaches. One shared digital environment. Individual decisions, collective consequences.

The Data We're Ignoring

More in Common research shows that parental concern about digital risks is far from universal. Many parents aren't "diligent or, as the research shows, all that concerned about the mental and physical risks posed to young people when they go online."

This creates an impossible situation: Conscientious parents cannot compensate for those who aren't concerned. Careful oversight in one home doesn't counteract unrestricted access in another when children attend school together.

Nicholson writes: "We can be pro-tech and also pro-safety, but we have to be able to talk to each other and come to some agreement around what we, as a country, will allow for our children."

That word "we" is the part we keep avoiding.

Why Neither Australia Nor America Got This Right

Australia's approach: Legislate the collective action problem away. Remove the burden from individual parents by making platforms enforce age restrictions.

Result: Technical bypasses available immediately. VPNs cost less than £20/month, not the "thousands of dollars" claimed by eSafety Commissioner Julie Inman Grant. Age verification relies on behavioral profiling that raises serious privacy concerns.

America's approach: Rely on individual parental choice. Provide tools (parental controls, platform safety features) and trust families to make appropriate decisions.

Result: A decade of increasing adolescent mental health crisis. Platforms optimized for engagement over safety. Individual parents fighting alone against billion-dollar behavior modification systems.

Neither works because both avoid the fundamental problem: Individual solutions cannot solve collective action problems.

What Collective Action Could Look Like

We have examples of this working in other domains:

Seatbelt laws didn't rely on individual parents deciding whether to buckle their children. We established collective standards enforced through regulation.

School start times are set by districts, not negotiated family by family, because education requires coordination.

Digital safety could work the same way:

School-wide agreements on phone policies that apply to all students, removing the "everyone else has one" argument.

Community norms around social media age thresholds, making it easier for individual parents to enforce delays.

Coordinated pressure on platforms for meaningful safety-by-design rather than fragmented individual complaints that companies can ignore.

Local coalitions of parents establishing shared expectations, so no family is fighting alone.

The Question Nobody Wants to Answer

Here's the uncomfortable part:

Collective action requires giving up some individual autonomy.

It means agreeing that just because you think your child is ready for unrestricted Instagram access at age 10 doesn't mean you should provide it, because your choice affects other children.

It means accepting that community standards might be stricter than your personal risk tolerance, or more permissive.

It means prioritising collective child wellbeing over individual parental prerogative.

For a culture deeply committed to individual choice, this feels un-American. Authoritarian, even.

But we're already making this trade-off. The current system prioritises individual parent choice over collective child safety - and the results are clear.

Adolescent mental health has deteriorated significantly over the past 15 years, coinciding precisely with smartphone and social media adoption.

The Path Forward

Nicholson's argument isn't that legislation is the answer. Australia's failed implementation proves it's not that simple.

Her argument is that we need to move beyond individual action as the primary or only strategy.

That might mean:

• Parent coalitions that establish shared norms

• School district policies that apply universally

• State / District level regulation that platforms must follow

• Federal standards for design features proven harmful to children

It definitely means having conversations with other parents in your community about what collective standards could look like.

Those conversations are hard. They require navigating different parenting philosophies, risk tolerances, and values. They require admitting that your individual choices affect other people's children.

But the alternative is what we have now: A decade of individual parents fighting billion-dollar platforms alone. Conscientious families bearing the social cost of setting boundaries their children's peers don't have. Mental health outcomes getting worse while we wait for individual parenting to solve a structural problem.

A Decade of Failure Should Change Our Strategy

The individual parenting approach has had a fair trial. The results are in.

Screen time limits in one household don't protect that child when their entire social world organises around platforms with no time limits.

Parental controls on one device don't help when group chats happen on unmonitored ones.

Age-appropriate content in one home doesn't shield children from what their classmates share at school.

Individual action cannot solve collective action problems. This is structural, not moral.

The question isn't whether collective approaches infringe on parental autonomy. The question is whether we're willing to continue the current approach while adolescent mental health deteriorates.

How much worse do the outcomes need to get before we try something different?

What’s your experience been? Are you fighting this battle alone in your community, or have you found ways to build collective standards?

A Tale of Two Cities

A school wants to buy an AI system that will automatically grade student essays and predict which children might drop out. It’s real technology that’s being sold to schools right now, and it promises to save teachers time while identifying struggling students early.

In Frankfurt, this is what happens next.

Before the school can even consider purchasing the system, the company selling it must prove something fundamental: that their AI has undergone mandatory safety testing. Not voluntary. Not “we promise it’s fine.” Mandatory. The company must demonstrate that their system has a risk management process, that the data used to train it is representative and free from errors, and that teachers will maintain oversight of every decision the AI makes. They must provide technical documentation showing all of this, register the system in an EU database, and design it so that students can challenge any decision it makes about them.

Then the school must conduct its own assessment. They must evaluate how this system might affect children’s fundamental rights. They must inform parents that AI is being used. They must ensure that no automated decision about a child’s education happens without a human being involved. They must keep detailed logs of what the AI does.

If either the company or the school fails to do any of this, they face fines of up to thirty-five million euros or seven percent of global annual turnover, whichever is higher.

Now let’s cross the Channel to Birmingham.

The headteacher receives the same proposal from the same company. The decision process looks like this: Does it fit the budget? Does it sound good? Buy it. That’s it.

There’s no mandatory safety assessment. No requirement to tell parents. No standards the system must meet. No penalties for getting it wrong unless it violates existing data protection law, which, according to research from the London School of Economics, isn’t being properly enforced in schools anyway. The headteacher might check if the school has an AI policy, but there’s a fifty-fifty chance one doesn’t exist. Half of UK schools have no AI policy whatsoever.

This is the reality of AI in education right now. Two countries separated by a short train journey under the sea, and two completely different approaches to protecting children.

How the EU Decided What’s Too Dangerous

The European Union passed the world’s first comprehensive AI law in 2024, and it came into full force this year. It’s built around a simple idea: the riskier the AI system, the stricter the rules. At the top of the risk pyramid are AI applications that are simply banned outright. Things like social credit scoring systems, or AI that tries to manipulate people without them knowing, or systems that scrape facial images from the internet to build databases. These are considered so dangerous that they’re prohibited entirely.

Then there’s emotion recognition AI. You might have seen marketing materials for systems that claim to detect whether students are engaged by reading their facial expressions. Some vendors pitch these as attention-monitoring tools for online learning. In the EU, using this technology in schools is banned. There’s an exception for genuine medical or safety purposes, but the default position is clear: schools cannot deploy AI to read children’s emotions. The science behind these systems is questionable, the potential for discrimination is significant, and the fundamental rights implications are serious enough that the EU decided they simply shouldn’t be used on children in educational settings.

In the UK, companies can market these systems to schools tomorrow. There’s nothing stopping them.

When AI Grades Your Child’s Work

Below the banned category sits “high-risk” AI, and this is where most educational technology lands. The EU’s legislation is explicit about this. If an AI system is used to determine who gets into which school or university, it’s high-risk. If it’s used to grade students or evaluate their work, it’s high-risk. If it assesses what level of education a child should receive, it’s high-risk. If it monitors students during exams to detect cheating, it’s high-risk.

There’s also a crucial catch-all provision. If an AI system profiles students—meaning it analyses personal data to evaluate their traits, predict their behaviour, or assess their performance—it’s automatically classified as high-risk, even if the company making it thinks it’s perfectly safe. The mere fact that it’s profiling children triggers the classification.

What does high-risk classification mean in practice? It means the companies building these systems must meet specific technical standards. They must test for bias and discrimination. They must ensure their AI doesn’t make decisions based on characteristics like race, gender, or socioeconomic background. They must design their systems so that humans remain in control, not just present but actively overseeing and able to override any decision the AI makes. They must maintain detailed documentation proving all of this, and schools must be able to show regulators that they’re using these systems responsibly.

The companies that make tools like Turnitin, which now uses AI to detect AI-generated writing and assist with grading, or Gradescope, which uses AI to group similar answers and speed up marking, would need to meet these standards in the EU. The proctoring software that became common during the pandemic, the kind that watches students through their webcams and flags “suspicious” behaviour, would need to meet these standards. The algorithms that universities increasingly use to screen applications and predict which students will succeed would need to meet these standards.

In the UK, none of this is required. These tools are being deployed in schools and universities right now with no mandatory safety assessment, no required bias testing, no mandated human oversight, and no obligation to inform parents that their children’s education is being shaped by algorithms.

The Tools Already in Your Child’s Classroom

Now here’s what makes this particularly stark: we know exactly what tools are being used in UK classrooms because teachers have been surveyed extensively. 85% percent of teachers are now using AI tools, according to research by the Center for Democracy and Technology. The explosion in adoption has been dramatic and rapid.

Let’s talk about what that actually means. Take Turnitin, which is widely used across UK schools and universities. Originally a plagiarism detection tool, it now uses AI to detect AI-generated writing and to assist teachers with grading. Under EU law, because it evaluates student work, it would be classified as high-risk. It would need to prove its detection systems don’t discriminate against students whose first language isn’t English, which is a documented problem with AI writing detection. It would need to demonstrate that its grading assistance features are accurate and don’t introduce bias. Schools would need to ensure teachers maintain oversight and that students can challenge its assessments.

Or consider the proctoring software that became ubiquitous during pandemic-era remote learning. Systems like Proctorio watch students through their webcams during exams, track their eye movements, monitor whether they look away from the screen, flag them if they appear to be talking to someone, and generate suspicion scores based on their behaviour. Some of these systems claim to detect emotional states. They’re monitoring children for prohibited behaviour during tests, which puts them squarely in the high-risk category. The ones attempting emotion recognition might be partially or fully banned. Under EU law, these would require extensive safeguards, transparency, and the ability for students to contest their flagging decisions. In the UK, they simply require a school to buy a license.

Even more concerning are the tools being used at the admissions level. Universities increasingly use AI to screen applications, assess personal statements, and predict which students are likely to succeed. These systems make decisions about who gets access to education, which is explicitly listed as high-risk in the EU framework. They require bias testing, transparency about how decisions are made, human oversight, and the right for applicants to challenge decisions. In the UK, these tools are proliferating with no such requirements. An algorithm could reject your child’s university application, and you might never know it was an algorithm that made the decision.

AI Ethics

Ofqual, which regulates qualifications in England, has explicitly warned that using AI as the sole method for marking student work is unlawful. Yet according to teacher surveys, it’s happening. Some teachers report that AI tools are generating entire assessment rubrics, grading student work, and providing feedback with minimal human review. The time-saving benefits are significant, which is why teachers are doing it. But without requirements for bias testing, accuracy verification, or mandatory human oversight, there’s no way to know if these systems are fair, consistent, or educationally sound.

Here’s what we’re not seeing in the UK: evidence that any of this actually improves education. Research from the London School of Economics notes that there is “no independent research-based evidence on claimed benefits” of AI in education. We’re deploying these tools at scale, transforming how children learn and how their work is assessed, and we have no rigorous research proving it makes things better. We’re making a massive bet on technology that hasn’t been proven to work.

Meanwhile, the concerns are mounting. Seventy percent of teachers, according to multiple surveys, worry that AI is making students dependent on technology for basic tasks and weakening their critical thinking skills. Sixty-eight percent of parents believe AI-assisted cheating is at least somewhat common. AI detection tools are disproportionately flagging work by students whose first language isn’t English, effectively accusing them of cheating when they’ve done nothing wrong. There’s no requirement to test for this bias before deploying these tools.

Why Britain Chose to Look The Other Away

So why has the UK chosen this path? The government’s position is explicit. They call it maintaining “our pro-innovation advantage.” They want the UK to be an attractive place for AI companies to invest and develop their technology. In January of this year, the government released something called the AI Opportunities Action Plan, which sets out its vision for making Britain a global AI leader. It talks about AI growth zones, increased compute infrastructure, and removing regulatory barriers. It emphasises that Britain’s light-touch approach to regulation is “a source of strength” and that the government must “be careful to preserve this.”

The subtext is clear: regulations might scare tech companies away. If we impose safety requirements, if we demand testing and transparency, if we fine companies for getting things wrong, they might choose to invest elsewhere. So the government has decided that keeping tech companies happy is more important than protecting children’s education and data.

The first dedicated UK AI legislation isn’t expected before the second half of 2026 at the earliest, and when it does arrive, it’s only going to target what the government calls “the handful of leading AI companies developing the most powerful models.” School-level AI deployment will likely remain unregulated. The government has five principles—safety, transparency, fairness, accountability, and contest-ability—but they’re not legally binding. They’re guidance. Regulators like the Information Commissioner’s Office and Ofcom are encouraged to apply these principles in their sectors, but there are no penalties for schools or companies that ignore them. It’s a voluntary framework masquerading as governance.

This is what an unregulated market looks like. Well-intentioned teachers trying to save time, tech companies eager to sell their products, schools under pressure to adopt “innovative” solutions, and children caught in the middle of an experiment nobody designed.

What Good AI Policies Actually Look Like in UK Schools

Most schools are making this up as they go along, and most parents don’t even know it’s happening. So let me share what effective AI policies actually look like in practice.

Here’s the shocking truth: by December 2024, nearly 70% of UK pupils were using AI tools, yet only one in four teachers uses AI daily. Even more concerning, 22% of secondary school pupils don’t understand what AI even means, and 43% can’t recognise the risks. We’ve got children navigating powerful technology without proper guidance whilst many schools scramble to catch up.

The Department for Education’s recent Ofsted study interviewed 21 “early adopter” schools across England who’ve been using AI for at least 12 months. Here’s what actually works:

1. They Appointed AI Champions

Nearly all successful schools designated an “AI champion” – typically a tech-savvy teacher who demystifies AI for colleagues and demonstrates practical applications. These champions play crucial roles in building staff confidence and showing teachers how AI can specifically help with their subject needs. As one school leader put it: “The biggest risk is doing nothing.”

2. They Established Clear Governance Structures

The most effective schools assigned three distinct responsibilities:

• Strategic oversight (usually a governor or senior leader)

• Operational management (digital learning lead or head of teaching and learning)

• Subject-specific guidance (heads of department adapting policies to their contexts)

Many schools created AI Ethics Groups comprising Teaching and Learning, IT Systems, and GDPR teams that meet monthly to ensure AI applications align with school strategy and ethical standards.

3. They Prioritised Teacher Workload Reduction

Schools focused primarily on using AI to reduce teacher workload – saving teachers an average of 5.1 hours per week through:

• Lesson planning and resource creation

• Administrative tasks and parent communications

• Generating differentiated materials

• Drafting rubrics and assessment criteria

Critically, this freed up teachers to focus on what matters most: face-to-face teaching and building student relationships.

SWGfL’s comprehensive policy template (South West Grid for Learning) identifies these crucial elements:

Data Protection & Privacy:

• Staff and pupils must NEVER enter personal information into AI tools

• Schools must comply with UK GDPR regulations

• Only school-approved AI tools with verified data security should be used

• All AI tools must be vetted before deployment

Safeguarding Requirements:

• Schools must enforce age restrictions (many AI tools are 18+)

• Close supervision required for any student AI use

• Designated Safeguarding Lead (DSL) oversees AI’s impact on student safety

• Updated policies must address AI-generated deepfakes, grooming risks, and peer-on-peer abuse

Assessment & Academic Integrity:

• Clear boundaries for different assessment contexts

• Policies must distinguish between genuine confusion, poor judgment, and intentional deception

• Follow Joint Council for Qualifications (JCQ) guidance on AI in controlled assessments

• Schools should document cases to refine policies (without naming individuals)

Transparency & Training:

• All AI-generated or AI-assisted content must be clearly labelled

• Staff receive training on advantages, risks, and ethical considerations

• Students taught to critically evaluate AI-generated content

• Parents informed through newsletters and school sessions

What This Means in Practice: Basingstoke College Case Study

Basingstoke College of Technology provides a brilliant example. They:

1. Created phased implementation (February-December) with controlled prototypes

2. Developed an AI Misuse Policy alongside the main policy

3. Achieved 5.1 hours per week time savings for teachers

4. Focused on developing students’ human skills alongside AI literacy

5. Established monthly AI Ethics Group meetings

Their approach? Use AI to handle administrative burden so teachers can focus on developing critical thinking, creativity, and interpersonal skills – the very capabilities AI can’t replicate.

The Critical Safeguarding Gaps Schools Must Address

Here’s what keeps safeguarding experts up at night:

Deepfake Risks: Multiple incidents worldwide show students creating fake images/videos that cause distress to peers and teachers. Schools need clear reporting procedures and response protocols.

Sextortion: The FBI issued joint warnings with UK law enforcement about criminals using AI-generated deepfakes to extort children, particularly boys.

Privacy Violations: Teachers’ social media photos being scraped to create AI images without consent. Schools must advise staff on protecting their online presence.

Age Verification Failures: Most generative AI tools have 18+ terms of service, yet children routinely access them. Schools must enforce these restrictions.

What Parents Should Ask Their School

Don’t assume your school has this sorted. Here are the questions you should be asking:

1. Do you have a written AI policy? (If not, that’s a red flag)

2. Who is your designated AI lead or champion?

3. What AI tools are approved for staff use? For student use?

4. How do you enforce age restrictions on AI tools?

5. What training have teachers received on AI risks and safeguarding?

6. How are you teaching students about AI literacy and critical evaluation?

7. What’s your process for vetting new AI tools before approval?

8. How do you monitor for AI misuse or deepfake creation?

The Bottom Line

The best school AI policies don’t ban technology, they govern it intelligently. They:

• Reduce teacher workload so educators can focus on human connection

• Protect student data and privacy rigorously

• Teach critical AI literacy as a core skill

• Maintain academic integrity without paranoia

• Update regularly as technology evolves

• Keep safeguarding at the absolute centre

As Government guidance now states: “Schools should consider online safety, including AI, when creating and implementing their approach to safeguarding.” This isn’t optional anymore – it’s statutory.

The challenge for parents? Most schools are still in the early experimental phase. The technology is advancing faster than educational policy can keep up. That means parents need to be informed advocates, asking the right questions and ensuring their children’s schools take this seriously.

Because here’s the truth: children are already using AI. The question isn’t whether to allow it, but how to govern it responsibly whilst we still have a chance to shape healthy digital habits.

For years, parents have been told to limit "screen time." The advice is everywhere: from paediatricians, from schools, from parenting books. Reduce screens. Set time limits. Balance digital with physical.

But that advice has always been frustratingly imprecise. Because "screens" covers an enormous range of activities. Reading an ebook is screens. Playing chess against a computer is screens. Watching David Attenborough documentaries is screens. Scrolling TikTok is screens.

Are those really equivalent?

A new study from Sweden's Karolinska Institutet and Oregon Health & Science University suggests they're not. And the differences matter enormously.

The Study Design

The researchers tracked 8,324 children over four years, from ages 9-10 to age 14. This wasn't a survey where parents estimated screen time. This was longitudinal tracking of actual daily habits.

The children's activities were categorised into three types:

Social media platforms: Instagram, Snapchat, TikTok, Facebook, Twitter, Messenger

Video games: any gaming activity

Television: passive viewing of scheduled or streaming content

The researchers measured attention levels and what they termed "inattention symptoms"—essentially, ADHD-like behaviours including difficulty focusing on tasks, getting easily distracted, and struggling to complete activities requiring sustained concentration.

The hypothesis: Sweden, like many countries, has seen a substantial rise in ADHD diagnoses over the past 15 years. That rise coincides almost exactly with the explosion of smartphone ownership and social media use among young people. Is there a connection?

The Finding

Children who spent significant time on social media showed a gradual, notable erosion of their ability to concentrate.

That's the headline. But here's the critical bit that most reporting misses:

Children whose screen time was primarily gaming or television did NOT show the same pattern.

This is not about screens damaging brains. This is about social media specifically.

Why Social Media Is Different

Let's talk about what makes social media architecturally distinct from gaming or television.

Gaming has endpoints. You complete a level. You finish a match. You beat the boss. There are defined objectives and clear stopping points. The experience is bounded.

Television has schedules. Even with streaming, episodes have credits. Films have endings. There are natural breaks in the content flow where you make a conscious decision to continue or stop.

Social media has neither. There is no endpoint to your Instagram feed. There is no final TikTok. The algorithm continuously serves content, endlessly, calibrated precisely to your engagement patterns. Stopping requires active willpower to interrupt the flow, not passive acceptance of a natural boundary.

This architectural difference creates different cognitive effects.

When you're gaming, your attention is directed towards achieving goals. The feedback loops are clear: you succeed or fail at defined tasks. Your brain is problem-solving, strategising, executing plans.

When you're watching television, your attention is passive. You're receiving information, not generating it. There's no requirement for constant interaction or decision-making.

When you're on social media, your attention is being actively manipulated by systems designed to maximise engagement. Every scroll is a micro-decision. Every post is evaluated for relevance. Every notification is an interruption. The cognitive load is constant and diffuse—you're neither focused on solving problems nor fully relaxed in passive consumption.

You're in a state of continuous partial attention.

And that state, maintained over hours daily across years, appears to degrade the capacity for sustained concentration.

The Usage Pattern Problem

Here's what the Swedish researchers noted: the average time children spent on social media platforms grew substantially during the study period.

This is important because most major social media platforms set their minimum age requirement at 13. Theoretically, none of the children should have had access to these platforms when the study began at ages 9-10.

But they did. And their usage increased as they got older.

This isn't a hypothetical risk or a future concern. This is happening now, to children below the stated age limits, on platforms that claim to verify age but manifestly do not.

The UK data supports this. According to Ofcom's 2025 analysis, 37% of 3-5 year-olds—children who cannot yet read—are using at least one social media platform. By age 11, over half of children have accounts on platforms officially restricted to 13+.

The enforcement mechanisms do not work. Which means the harms documented in this research are affecting far more children, far younger, than the platforms' age policies would suggest.

What "Inattention Symptoms" Actually Means

The study measured ADHD-like behaviours. That terminology is important because it's not claiming social media causes clinical ADHD—a neurological condition with genetic and developmental components.

What it's documenting is that children who use social media extensively display symptoms that look like ADHD: difficulty sustaining attention on tasks, increased distractibility, problems completing activities that require mental effort, and challenges with organisation and follow-through.

Whether this represents actual changes to brain structure and function or learned behavioural patterns is still being studied. The researchers plan to continue following these children beyond age 14 to track long-term effects.

But from a practical parenting perspective, the distinction may not matter much. If your child cannot focus on homework, cannot read for sustained periods, cannot have conversations without checking their phone, cannot complete tasks without distraction—whether that's neurological damage or learned behaviour, the functional impact is the same.

Their capacity for sustained attention is degraded.

The Gaming Exception

The finding that gaming doesn't show the same cognitive effects is likely to surprise many parents. Gaming has been demonised in parenting advice for decades. "Too much gaming" is a standard concern.

But this research suggests that concern may be misplaced, or at least imprecise.

Gaming requires active engagement. It requires sustained attention on specific objectives. It requires problem-solving and strategic thinking. From a cognitive perspective, gaming more closely resembles work than it resembles social media scrolling.

This doesn't mean unlimited gaming is beneficial. The study wasn't measuring whether gaming improves concentration, just that it doesn't degrade it the way social media does.

There are other concerns with excessive gaming: physical inactivity, social isolation, sleep disruption if gaming occurs late at night. But the specific cognitive effect of concentration degradation appears to be a social media problem, not a gaming problem.

What This Means Practically

If you're a parent trying to manage your child's digital life, this research suggests you should stop treating all screen time as equivalent.

The advice to "limit screens to two hours daily" conflates activities with very different cognitive effects. Two hours of Minecraft is not the same as two hours of Instagram.

What matters isn't screen time duration. What matters is what the screen is being used for.

Active engagement with defined goals (gaming, creative software, educational programs) appears to be cognitively neutral or potentially beneficial.

Passive consumption of finite content (television, films, YouTube videos with clear endpoints) appears to be cognitively neutral, though possibly displacing more beneficial activities.

Algorithm-driven infinite-scroll platforms designed to maximise engagement (Instagram, TikTok, Snapchat, Facebook) appear to degrade sustained attention capacity over time.

The first category can be permitted relatively freely. The second requires moderation to ensure balance. The third requires strict limitation or elimination.

The Implementation Challenge

Knowing social media is the problem doesn't make controlling access easier. Because by age 11, most UK children have smartphones. And smartphones provide access to social media regardless of parental rules or platform age restrictions.

You can delete Instagram from your child's phone. They can reinstall it via a browser. You can block social media on your home WiFi. They can use mobile data or a friend's WiFi. You can implement parental controls. They can use VPNs or find workarounds.

The Swedish researchers deliberately investigated this question because they wanted evidence to inform policy decisions. Their findings support Australia's approach of requiring platforms themselves to prevent underage access, rather than relying on parents to enforce restrictions on technology their children often understand better than they do.

But as we covered in last week's article on Australia's social media ban, enforcement at the platform level faces its own challenges. Age verification is difficult. VPNs exist. Fake accounts are trivial to create.

There is no perfect technical solution.

Which leaves parents in the familiar position of needing to set boundaries without perfect enforcement mechanisms, implement rules that children will test and circumvent, and have ongoing conversations about why these restrictions exist.

The Researcher's Hope

Samson Nivins, the study's first author, was explicit about the goal: "We hope that our findings will help parents and policymakers make well-informed decisions on healthy digital consumption that support children's cognitive development."

Well-informed decisions require understanding which digital activities cause harm and which don't.

For too long, the advice has been imprecise: reduce screens, limit technology, encourage offline activities. That advice is true but useless in practice because it doesn't help parents distinguish between Duolingo and TikTok, between creative video editing and passive scrolling, between problem-solving games and algorithm-driven engagement traps.

This research provides the distinction that practical parenting requires.

Social media specifically, not screens generally, degrades concentration in developing brains.

Not recommended. Not advised. Enforced, with fines up to 49.5 million Australian dollars (£26 million) for platforms that fail to comply.

The affected platforms are: Facebook, Instagram, TikTok, Snapchat, X (formerly Twitter), YouTube, Reddit, Twitch, Threads, and Kick.

Exempt platforms include: WhatsApp, YouTube Kids, educational services like Google Classroom, gaming platforms like Roblox, and messaging services used for healthcare.

And here's what makes this unprecedented: parents cannot give consent. There is no "parental permission" override. If you're under 16 in Australia, you cannot legally have an account on these platforms, regardless of what your parents think.

This is the world's largest experiment in social media restriction. And it's happening right now.

What's Happening This Week

Meta began removing under-16 accounts on 4th December—six days before the law takes effect. Instagram alone has approximately 350,000 Australian users aged 13-15. All of those accounts are being deleted.

Users can prove their age by scanning their face or providing identity documents. If they can't prove they're 16 or older, they're locked out. Their content isn't deleted—it's waiting for them when they turn 16—but they cannot access it now.

YouTube is taking a different approach. Rather than requiring accounts, they're allowing under-16s to use the platform without logging in. Which means those users lose access to all the safety features, parental controls, and content filters that only work for logged-in accounts.

YouTube's official position: "This law will make Australian kids less safe on YouTube."

The Australian Communications Minister's response: "If YouTube is reminding us that there's content not appropriate for age-restricted users on their website, that's a problem YouTube needs to fix."

This is the tension at the heart of the implementation: platforms designed the safety features for logged-in users. The law requires children not to have accounts. Therefore, children lose the safety features.

Nobody designed for this scenario.

The Numbers Tell Two Stories

70% of Australian voters support the ban. That’s a supermajority, overwhelming public backing.

But 58% of those same voters don’t think it will work.

Think about what that means. People want this. People support this. But people don’t believe in it.

This isn’t contradictory—it’s rational. Parents want someone to do something about the documented harms of social media on children. But parents also understand their children. They understand that age verification can be circumvented. They understand that teenagers are more technically competent than legislators assume.

The polling shows that 53% of parents plan to “pick and choose” which platforms to allow their children to use, rather than enforcing full compliance. 29% intend full compliance. 13% plan to take no action at all.

So even with legal force behind the ban, actual enforcement will be inconsistent, variable, and dependent on individual family decisions.

Which raises the question: what’s the point of a law that most people intend to selectively follow?

The Enforcement Problem

Here’s the technical challenge: how do you verify someone’s age online?

The Australian law doesn’t specify. It requires platforms to take “reasonable steps” but doesn’t define what those steps are.

Meta’s approach: facial scanning or ID upload.

The problems with this:

1. Facial age estimation has error rates, particularly for young people whose faces are still developing

2. Not all 16-year-olds have government ID

3. Uploading ID to social media platforms raises massive privacy concerns

4. VPNs exist—Australian teenagers can simply appear to be from another country

The eSafety Commissioner acknowledges this is “a complex task” and is “consulting with platforms about effective methods.”

Translation: they’re making this up as they go along.

To be clear, that’s not a criticism. This is genuinely novel regulatory territory. No country has tried this before. There is no playbook.

But it does mean that December 10th isn’t the finish line—it’s the starting line for a years-long process of refinement, circumvention, counter-measures, and adaptation.

What Happens to Children?

The research on social media harm is compelling. The Swedish study from December 2025 tracked 8,324 children and found that social media use specifically—not gaming, not television—was associated with declining concentration. The Children’s Hospital Philadelphia study found smartphone ownership by age 12 predicted worse mental health outcomes.

The Australian government cites this evidence. They argue that protecting children at a “critical stage of development” justifies the restriction.

But here’s what the research doesn’t tell us: what happens when you remove social media from teenagers’ lives?

Do they develop better? Or do they migrate to less regulated platforms where the harms are worse?

Dr Brittany Ferdinands from the University of Sydney warns: “Preventing under-16s from having social media accounts won’t necessarily stop them using them. In fact, it may push their activity underground.”

This is the unintended consequence problem. When you ban something, you don’t eliminate it—you make it harder to monitor.

Under-16s who currently use Instagram do so visibly. Parents can see the app on their phone. Schools can discuss it. Friends can report concerning behaviour.

Under-16s who use Instagram via a VPN, with a fake age, claiming to be from another country, do so invisibly.

Which scenario is safer?

The UK Context

Lord Nash proposed an identical ban for the UK in June 2025, as an amendment to the Children’s Wellbeing and Schools Bill. It cited Australia’s “pioneering model” as evidence that this approach works.

Except we don’t know if it works yet. Australia’s law takes effect today.

A Westminster Hall debate is scheduled for 15th December—five days after Australia’s implementation begins. The UK will be watching Australia’s first week to inform their own policy discussions.

An e-petition calling for a UK under-16 ban has received over 127,000 signatures. That’s enough to trigger parliamentary consideration.

The UK government’s current position: “The government is not currently minded to support a ban for children under 16.” Instead, they’re focusing on implementing the Online Safety Act 2023, which requires platforms to protect children through design changes rather than blanket age restrictions.

But that position may change based on what happens in Australia this month.

What Actually Works

Here’s what we know from Ofcom data: almost three-quarters of UK teenagers aged 13-17 have encountered one or more potential harms online. Three in five secondary school-aged children have been contacted online in ways that made them uncomfortable.

The Online Safety Act approach targets this through:

• Algorithmic filtering to prevent harmful content reaching children

• Robust age verification for the most harmful content (pornography, self-harm material)

• Mandatory content moderation systems

• Transparent reporting mechanisms

Does this work better than a blanket ban? We don’t know yet. The Online Safety Act’s children’s safety codes only came into effect in July 2025.

So we’re running two simultaneous experiments:

• Australia: blanket age restriction

• UK: platform responsibility with design requirements

Both are untested. Both are unprecedented. Both may fail.

What Parents Can Do

If you’re waiting for government policy to protect your children online, you’re waiting for an experiment to conclude.

That experiment will take years. Your children are online now.

The evidence is clear on certain points:

1. Early smartphone ownership (before age 12) is associated with worse outcomes

2. Social media use specifically (not all screen time) harms concentration

3. Platform features like disappearing messages facilitate grooming

4. Younger children are at higher risk than older adolescents

But evidence on blanket bans is non-existent because no one has tried it at scale before.

What this means practically:

Delay smartphone ownership - The Children’s Hospital Philadelphia data shows effects are measurable and significant. Age 12 is too young for most children. Age 14 or 15 may be more appropriate.

Distinguish between platforms - The Swedish research shows social media is uniquely problematic. Gaming and video watching show different patterns. “No screens” is less precise than “no Instagram.”

Understand specific features - As the Snapchat data shows, 48% of grooming happens on one platform because of architectural features. Knowing how Quick Add or Snap Map work matters more than general “be careful” advice.

Accept that supervision is necessary - 70% of Australians support a ban because they want help. You cannot outsource this to platforms or governments. The tools don’t exist yet.

The Bottom Line

Australia’s ban takes effect today. We’ll know within weeks whether it’s enforceable. We’ll know within months whether children circumvent it. We’ll know within years whether it actually improves outcomes.

Until then, this is the world’s most expensive research project in child online safety.

The UK is watching, Parents everywhere are watching.

And hundreds of thousands of Australian teenagers are about to lose access to the platforms they’ve used daily for years.

What happens next will inform policy globally.

But it won’t inform your decisions today.

Because those decisions can’t wait for experimental results.

In the year 2024/25, UK police forces recorded 7,263 Sexual Communication with a Child offences. That’s nearly double the number recorded when the offence was first introduced in 2017/18.

Of the 2,111 cases where police could identify which platform was used, 48% occurred on Snapchat. That’s 1,013 children, groomed on a single platform.

For context: WhatsApp and Facebook/Instagram each accounted for 9%. TikTok, despite its massive youth audience, represented a smaller percentage still.

The question is: why Snapchat?

The answer is architecture. Snapchat’s core features—the ones that define the platform, the ones that made it popular—create what engineers might call “optimal conditions” for predatory behaviour.

This isn’t about blaming technology. This is about understanding it. Because if you don’t understand how these systems work, you can’t protect your children from them.

The Fundamental Problem: Ephemeral by Design

Snapchat’s founding principle was ephemerality. Messages disappear. Photos vanish. Stories expire after 24 hours.

In 2011, when Snapchat launched, this was positioned as privacy. And technically, it is. But privacy cuts both ways.

When a message disappears, two things happen:

1. The recipient can’t review what was said

2. There’s no evidence trail for parents or authorities

From a predator’s perspective, this is ideal. Grooming is a process—it’s gradual, incremental, testing boundaries over time. Disappearing messages mean there’s no record of that progression. No evidence of the early conversations where things were innocent. No proof of when the tone shifted.

The NSPCC data shows that 80% of victims were girls, and the youngest recorded victim was just 4 years old. These are children who may not even understand what’s happening, much less know to screenshot evidence before it vanishes.

Feature One: Quick Add

Quick Add is Snapchat’s friend suggestion algorithm. It surfaces potential connections based on mutual friends, phone contacts, and location data.

Here’s how it works technically:

• Snapchat analyses your friend graph

• It identifies people who are 2-3 degrees of separation away

• It surfaces these suggestions prominently in the app

• Users can add these complete strangers with a single tap

For legitimate users, this helps you find friends of friends. For predators, this is a discovery mechanism.

A predator only needs to connect with one child in a community—perhaps through a fake account posing as another teenager. Once that connection is made, Quick Add will suggest dozens of other children from the same school, neighbourhood, or social circle.

The system is designed to maximise connections. It cannot distinguish between “teenager wants to find their friend’s friend” and “adult wants to access multiple children from the same community.”

And because Snapchat’s verification system is minimal, creating fake accounts is trivial. The youngest victim recorded was 4—an age at which children cannot meaningfully consent to being on the platform at all, let alone evaluate whether a new connection is legitimate.

Feature Two: Snap Map

Snap Map is a real-time location sharing feature. When enabled, it shows your precise location to your friends—or to everyone, if you’ve set it to public.

From an engineering perspective, this is impressive. Real-time geolocation, rendered on an interactive map, with user-generated content overlaid. Technically sophisticated.

From a safeguarding perspective, this is a disaster.

Consider the attack vector:

1. A predator adds a child via Quick Add

2. The child accepts, thinking it’s someone from school

3. Snap Map reveals the child’s location in real-time

4. The predator now knows when the child is home, when they’re at school, when they’re walking to the shops

The Greater Manchester Police case from December 2025 involved a Tameside man who used “fake online identities to groom and exploit young victims.” He was convicted of 30 offences. The investigation found “multiple email and social media accounts” and “archived WhatsApp chats with teenage girls.”

Snap Map doesn’t appear in that specific case detail, but it’s representative of the pattern: predators are using platform features designed for convenience to gather intelligence on children’s movements and routines.

Feature Three: Streaks

Streaks are a gamification feature. If you and a friend send snaps to each other for consecutive days, you build a “streak.” The number next to their name shows how many days you’ve maintained it.

This seems innocent. It’s just a number. But it creates a powerful psychological pressure, particularly for young people.

Maintaining a streak requires daily contact. If you break it, you lose progress. Many teenagers report feeling genuine anxiety about losing streaks, particularly with close friends.

For groomers, this is leverage.

Research from organisations like the Lucy Faithfull Foundation shows that groomers deliberately create dependency. They make children feel special, needed, obligated. Streaks provide a pre-built mechanism for this.

A predator can establish a streak, then use the child’s desire to maintain it as a reason for daily contact. “You don’t want to lose our streak, do you?” becomes a way to ensure regular communication, to normalise the relationship, to create expectations of daily interaction.

The feature is designed to increase engagement metrics. It succeeds.

But engagement with whom?

Feature Four: Private Stories

Stories on Snapchat are photos or videos visible for 24 hours. But unlike Instagram Stories, Snapchat allows “Private Stories”—content shared only with selected people.

Here’s the technical implementation:

• Users can create multiple stories with different audiences

• They can choose exactly who sees each story

• Recipients don’t know who else is on the list

• There’s no way for parents to see this content unless they’re specifically included

For teenagers, this is about social circles. One story for close friends, another for acquaintances, another for family.

For predators, this is about isolation.

The Gloucestershire Police case from November 2025 involved a man who groomed multiple children across different counties. He “would also film himself sexually abusing them, and groomed them into sending him videos of things he had asked them to do.”

This is the pattern: predators don’t just request images. They create conditions where children produce content specifically for them. Private Stories provide a mechanism for this—a way to share content with “just us,” away from parents, away from friends who might question what’s happening.

Feature Five: The Deletion Architecture

Here’s what happens technically when you send a Snap:

1. The message is encrypted and sent to Snapchat’s servers

2. The recipient is notified

3. They open it, view it for up to 10 seconds (depending on settings)

4. The message is marked as “viewed” and deleted from Snapchat’s servers

Except that’s not quite what happens.

Screenshots exist. Screen recording exists. Third-party apps that bypass Snapchat’s screenshot detection exist. The content isn’t really gone—it’s just gone from Snapchat’s official infrastructure.

But here’s the critical bit: children don’t know this. They believe the promise of ephemerality. They trust the disappearing message.

Predators know better.

The NSPCC data shows 7,263 recorded offences, but explicitly states the real number is “much higher due to abuse happening in private spaces where harms can be harder to detect.”

Disappearing messages make detection harder. They make evidence collection harder. They make intervention harder.

And when detection is harder, predators are safer.

Why This Matters More Than Content Moderation

You might be thinking: surely Snapchat moderates content? Surely there are reporting mechanisms?

Yes. There are. They’re industry-standard. They’re probably no worse than other platforms.

But content moderation is reactive. It depends on:

1. Harmful content being posted

2. Someone seeing it and reporting it

3. Moderators reviewing it quickly enough

4. Action being taken

Grooming doesn’t work like that.

Grooming happens in private messages. It’s conversational. It’s gradual. There’s nothing to moderate until it’s far too late—until explicit images have been exchanged, until abuse has occurred, until the damage is done.

By the time there’s something to report, the child has already been groomed. The features have already done their job.

This is why 48% of recorded offences happened on Snapchat. Not because Snapchat has worse moderation. But because Snapchat’s features create better conditions for grooming to succeed undetected.

What the Research Shows

The Swedish study released in December 2025 tracked 8,324 children from ages 9-10 to 14. It found that social media use specifically—not gaming, not television—was associated with declining concentration and increased ADHD-like symptoms.

The Children’s Hospital Philadelphia study from the same month analysed 10,000+ adolescents and found that merely owning a smartphone by age 12 increased risks of depression, poor sleep, and obesity.

These studies don’t mention Snapchat specifically. But they provide context.

Children’s brains are developing. They’re learning social skills, emotional regulation, risk assessment. They’re not equipped to evaluate the intentions of strangers or the implications of sharing their location or the permanence of supposedly temporary content.

And they’re certainly not equipped to understand the systematic ways that platform features can be weaponised against them.

The youngest victim is 4 years old

Let’s return to that statistic: the youngest victim of online grooming was a 4-year-old boy.

At 4, children are learning to count, to recognise letters, to share toys. They cannot read privacy policies. They cannot understand age restrictions. They cannot meaningfully consent to anything, let alone participation in platforms designed for teenagers.

And yet, 37% of UK 3-5 year-olds are on social media. That’s nearly 800,000 pre-schoolers, according to Ofcom data analysed by the Centre for Social Justice in December 2025.

These children aren’t circumventing parental controls or lying about their age. They’re being placed on these platforms by adults. By parents who perhaps think a few Disney videos are harmless. By older siblings who don’t understand the risk. By babysitters who need a distraction tool.

And once they’re there, the features don’t care about age. Quick Add will still suggest strangers. Snap Map will still broadcast location. The deletion architecture will still erase evidence.

The platform doesn’t ask “should a 4-year-old have access to this feature?” It just asks “is the feature enabled?”

What Parents Need to Know

You cannot rely on age restrictions. The 48% statistic exists despite Snapchat requiring users to be 13+.

You cannot rely on platform safety features. They’re reactive, not preventive.

You cannot rely on “having a conversation” being enough. Children lack the developmental capacity to assess these risks consistently.

What you can do:

Understand the features. Not just “Snapchat exists,” but how Quick Add works, what Snap Map reveals, why Streaks create pressure. You cannot discuss risks you don’t understand.

Check settings regularly. Snap Map can be set to Ghost Mode (hidden). Quick Add can be limited. But these settings reset with updates, or children change them.

Know that disappearing doesn’t mean gone. Teach your children that anything they send can be captured, saved, and shared, regardless of platform promises.

Recognise grooming patterns. Excessive attention, gifts, requests for secrecy, pressure to maintain contact. These patterns exist offline too—platforms just make them scale.

Accept that supervision is necessary. The data shows 7,263 recorded offences in a single year, double the number from six years ago. This isn’t rare. This isn’t “other people’s children.”

This is a statistical likelihood.

The Bottom Line

Snapchat isn’t uniquely evil. It’s uniquely effective—at engagement, at retention, at creating habitual use.

Those same features that make it effective for legitimate users make it effective for predators.

This is the engineering truth that nobody wants to hear: you cannot build systems that maximise connection and engagement without also maximising the risk of harmful connection and exploitative engagement.

The question isn’t whether platforms should exist. They will. They do.

The question is whether parents understand how they work.

Because 48% of online grooming happening on one platform isn’t random. It’s architecture.

And architecture can be understood.

A few months ago, the former infant school of one of my children posted on their school app warning parents about children as young as 5 recreating a violent "slapping game" they'd seen on Roblox during playground time.

At the time, I thought this was an isolated incident—perhaps a handful of children who'd somehow slipped through the parental controls. A concerning outlier, but surely not representative of what was happening across UK schools.

This week, new analysis from the Centre for Social Justice revealed that almost one million pre-school children in the UK are already active on social media platforms designed for teenagers and adults (Centreforsocialjustice) .

But it gets worse. We now know exactly which platforms these toddlers are on, how much time they're spending there, and what it's doing to their developing bodies and minds. The data doesn't just concern me as a parent—it should alarm anyone who cares about child development, public health, or the future generation growing up in front of screens.

Let me walk you through what the research actually shows, why it's happening, and what the physical evidence tells us about the cost we're paying for convenience.

Part One: The Platforms — Where Your Pre-Schooler Might Already Be

Let me be brutally specific about what the latest research reveals.

16% of 3-4 year-olds and 29% of 5-7 year-olds use TikTok (Statista) . Read that again. Nearly one in three children who've just started primary school—many who are still learning to read—are on TikTok, a platform with a minimum age requirement of 13.

But TikTok isn't working alone.

Among 5-7 year-olds specifically:

37% use WhatsApp, 30% use TikTok, 22% use Instagram, and 4% use Discord.

Nearly half of this age group—48% of children aged 5-7—now have their own personal profiles on YouTube or YouTube Kids.

Here's the part that should genuinely alarm every parent, educator, and policymaker reading this: a third (32%) of parents of 5-7 year-olds report that their child uses social media independently (Ofcom) .

Not with parental supervision.

Not co-viewing content together. Independently.

The year-on-year trajectory is even more concerning. 37% of parents of 3-5 year-olds now say their child uses social media, up from 29% in 2023. That's an increase of roughly 220,000 more pre-schoolers added to social media platforms in just one year.

Think about what this means in practical terms. Children who can't yet read fluently are scrolling through content designed to be addictive for adults. Toddlers who should be developing language skills through conversation and play are instead spending significant portions of their day staring at screens optimised by algorithms to maximise engagement—not development.

Why These Specific Platforms?

It's worth understanding why these particular platforms dominate among young children, because it tells us something important about how this happened.

YouTube/YouTube Kids leads the pack because parents often view it as "educational content." And to be fair, there is educational content on YouTube. But the algorithm doesn't distinguish between Peppa Pig and completely age-inappropriate material. Once a child is on the platform, they're one autoplay away from content that was never designed for their eyes.

WhatsApp has become ubiquitous because it's how families communicate, especially across borders or with grandparents. But 11% of 5-7 year-olds now have WhatsApp profiles (Ofcom) , which means they're not just receiving family messages—they're part of a messaging ecosystem that includes group chats, media sharing, and increasingly, contact with people outside their immediate family circle.

TikTok is the fastest-growing platform among young children because it's designed to be compulsively watchable. Short-form videos, instant gratification, infinite scroll. UK kids spend an average of 70 minutes per day on TikTok (Embryo) —and that's just one platform. The total daily screen time is significantly higher when you factor in YouTube, gaming, and other apps.

Instagram at 22% penetration among 5-7 year-olds represents something even more concerning: children this young are already being exposed to curated, filtered, performance-based social interaction. They're learning that life is something to be staged, photographed, and judged by others—before they've even developed a stable sense of self.

Part Two: What Social Media Is Stealing From Children’s Bodies

Here’s what haunts me most from the research, and what should be the headline that drives policy change:

“I’ve got two children [in my class] who physically cannot sit on the carpet”.

That’s a quote from a teacher, speaking to researchers about children in their classroom. Not children with diagnosed physical disabilities. Not children with identified medical conditions. Children whose core physical development has been so disrupted by excessive screen time that they lack the basic muscle strength to sit on the floor.

Let that sink in for a moment. Sitting on the floor is not an advanced physical skill. It’s something humans have done naturally for millennia. It requires core strength, yes, but it’s core strength that develops automatically through normal childhood movement: crawling, climbing, running, jumping, playing.

But normal childhood movement is being replaced by stillness. And stillness is stealing children’s physical development.

The Research Connects Clear Dots

Studies show that children exposed to over four hours of daily screen time appear more likely to experience fine motor and social-emotional skill delays. Fine motor skills aren’t abstract developmental milestones—they’re the abilities children need for basic independence: holding a pencil to write, buttoning their own coat, tying their shoelaces, using scissors, feeding themselves with utensils.

These are foundational skills. Without them, children struggle in school. They require more adult assistance. They fall behind peers in tasks that should be age-appropriate. And increasingly, teachers are reporting that these delays are becoming the norm rather than the exception.

The scale of screen time children are actually experiencing is staggering. UK children spend an average of 70 minutes per day on TikTok alone. That’s more than an hour on a single platform. It doesn’t include YouTube, which is the most popular service overall. It doesn’t include gaming, which has also seen significant increases. It doesn’t include WhatsApp messages or educational apps on tablets.

When you add it all up, many pre-school children are spending 3-4 hours or more per day on screens. A New Zealand study involving over 6,000 children aged 2-8 linked over 90 minutes of daily screen time to below-average performance in communication, writing and numeracy, with toddlers showing heightened behavioural issues and precursors to anxiety disorders.

And here’s the truly alarming part: 90 minutes is less than what many UK children are currently experiencing just on TikTok alone.

The Outdoor Play We’re Losing

There’s another side to this equation that’s equally important: what children aren’t doing while they’re on screens.

A UK survey by The National Trust found that children spend half as much time outside as their parents did Starglow Media. That’s a 50% reduction in outdoor play in just one generation.

Outdoor play is directly related to a child’s physical strength, weight and immune function. Kids who are active throughout childhood are usually more likely to engage in regular exercise later in life.

But beyond the general health benefits, outdoor play provides something screens fundamentally cannot: the development of spatial awareness, risk assessment, physical coordination, and gross motor skills that come from moving through three-dimensional space.

Playing outside allows kids to develop more advanced motor skills than children who spend most of their time indoors, including agility, balance and coordination. Kids who spend time playing outdoors are more likely to move in ways that challenge their muscles, bones and physical endurance.

When a child climbs a tree, they’re learning about their body’s capabilities and limits. They’re developing proprioception—the sense of where their body is in space. They’re building muscle strength, coordination, and confidence. They’re experiencing cause and effect in immediate, tangible ways.

When a child swipes through TikTok videos for 70 minutes, they’re developing... thumb dexterity. And algorithm-driven dopamine responses.

The trade-off couldn’t be starker.

Long-Term Physical Health Implications

The immediate developmental delays are concerning enough. But the long-term health trajectory is potentially catastrophic.

Studies show a strong link between high screen usage and childhood obesity, which increases the risk of diabetes and heart disease later in life. We’re not just talking about children who are a bit unfit or carry a few extra pounds. We’re talking about setting them on a path toward chronic illness before they’ve even finished primary school.

The generation of children growing up right now may be the first in modern history to have shorter life expectancies than their parents, largely due to obesity-related conditions. And excessive screen time—starting in the pre-school years—is a significant contributing factor.

Even 30-60 minutes of outdoor activity per day can improve attention, reduce anxiety, and encourage healthier lifestyle habits. But that requires actually prioritising outdoor time over screen time. And for hundreds of thousands of UK families, that’s not currently happening.

Part Three: Why This Is Happening—The Uncomfortable Truth

Lord Nash, former Minister at the Department for Education, was refreshingly direct in his assessment: “This research is deeply alarming. With hundreds of thousands of under-fives now on these platforms, children who haven’t yet learned to read, being fed content and algorithms designed to hook adults, should concern us all”.

But here’s what we need to be honest about: this isn’t happening because platforms have found a way to trick parents or bypass security measures. It’s happening because parents are actively creating accounts for children who are 3, 4, and 5 years old.

When researchers asked parents of children aged 5-7 about their awareness of minimum age requirements, they found that 30% would allow their child to have a profile on social media services before they had reached the minimum age required—up from 25% the previous year.

That’s not children lying about their age and sneaking onto platforms. That’s parents making a deliberate choice to give their pre-schoolers access to social media.

Understanding the Pressure Points

Before we judge too harshly, it’s worth understanding why parents are making these choices. The reasons are complex and often sympathetic:

Convenience and peace: Sometimes the only way to have 20 minutes to cook dinner is to hand a child a tablet with YouTube. The platform is incredibly effective at keeping children occupied and quiet.

Social pressure: “Everyone else’s child has one” is a powerful force, even for pre-schoolers. When other children in nursery or reception are talking about videos they’ve seen or games they play online, parents don’t want their child to feel excluded.

Educational justification: Many parents genuinely believe they’re giving their children educational content. YouTube is full of videos that claim to teach colours, numbers, letters, and problem-solving. The fact that these videos are embedded in a platform designed to maximise watch time gets overlooked.

Family connection: For families separated by distance—especially those with relatives abroad—platforms like WhatsApp feel like essential tools for maintaining relationships. The idea of excluding a 5-year-old from family video calls or photo sharing seems harsh.

Lack of awareness: Despite the research being available, many parents genuinely don’t know about the physical developmental impacts of excessive screen time. They see other children on devices and assume it must be fine.

The Platform Design Problem

But there’s another factor that doesn’t get discussed enough: these platforms are designed to be addictive. Not just for adults—for anyone with a nervous system capable of responding to intermittent rewards and dopamine hits.

TikTok’s algorithm is extraordinarily sophisticated. It learns what keeps a user watching within minutes and serves up more of exactly that content. For a pre-schooler, that might mean an endless stream of bright colours, loud sounds, fast movements, and simple entertainment. The algorithm doesn’t care about developmental appropriateness. It cares about engagement.

YouTube’s autoplay function means that even if a parent carefully selects an educational video for their child, the platform will automatically serve up whatever keeps the child watching next. And research has repeatedly shown that algorithms optimise for engagement, not education or wellbeing.

WhatsApp’s group chat dynamics and media sharing create social pressure even for young children to be available, responsive, and participatory. The same FOMO (fear of missing out) that affects adults affects children—arguably more acutely, because they haven’t yet developed the cognitive tools to recognize and resist it.

The Age Verification Problem

Creating an account for a 5-year-old on Instagram requires nothing more than typing in a birthdate that says they’re 13. There’s no verification. No checks. No barriers.

Platforms have known about this problem for years. They’ve done the bare minimum because meaningful age verification is expensive, technically challenging, and would significantly reduce their user numbers. Why invest millions in robust age verification when you can simply put the responsibility on parents to enforce age limits?

That’s finally changing—sort of. From 25 July 2025, platforms must use secure methods like facial scans, photo ID and credit card checks to verify ages for access to the most harmful material.

But notice that crucial qualifier: “the most harmful material.” This new requirement only applies to content like pornography, self-harm, suicide, and eating disorder material. Standard social media access for under-13s still relies on parental enforcement.

Which brings us back to the reality that 37% of parents of 3-5 year-olds report their child uses social media, and that number is growing year-on-year.

Part Four: What’s Changing (And What Isn’t)

There are some genuinely significant developments happening in UK online safety policy. Understanding what’s actually changing—and what remains problematic—is essential for parents trying to navigate this landscape.

The Online Safety Act 2023: Real Enforcement

The UK’s Online Safety Act is now in active enforcement, and it has teeth. Since March 2025, Ofcom has opened 21 investigations and issued major fines, including a £20,000 fine against 4chan for failing to respond to information requests.

Ofcom issued a £50,000 fine against the provider of a nudification site for failing to use age-checks to protect children from online pornography, and opened new investigations into five providers operating 20 pornography sites under its age assurance enforcement programme.

The maximum penalties are substantial: £18 million or 10% of worldwide revenue, whichever is greater. This isn’t theoretical anymore. Platforms are facing genuine financial consequences for failing to protect children.

From 25 July 2025, platforms hosting pornography or content encouraging self-harm, suicide, or eating disorders must implement robust age-checks using highly effective age assurance like facial age estimation, photo-ID matching, or credit card checks.

Children will also see fewer harmful posts and videos in their feeds, with platforms required to make sure their algorithms aren’t feeding children content that promotes harmful behaviours like bullying, hate speech, or dangerous online challenges.

This represents real progress. But there’s a critical gap.

The Under-13 Problem Remains

All of these enforcement actions and new requirements apply to platforms’ responsibilities once they know a child is using their service. But the fundamental problem remains: getting children onto platforms in the first place is still trivially easy.

For standard social media use—not the most harmful content, but everyday Instagram, TikTok, YouTube, WhatsApp—the barriers are minimal. Type in a fake birthdate. Create an account. Done.

And as we’ve seen, 30% of parents actively allow their children to have social media profiles before reaching minimum age requirements. The platforms aren’t circumventing parental controls. Parents are choosing not to use them.

The Australian Model: A More Radical Approach

Some countries are taking a dramatically different approach. In September 2025, Australia enacted legislation that mandates strict age verification to block under-16 access to social media, shielding kids from addictive algorithms and toxic content.

This is fundamentally different from the UK approach. Australia isn’t just requiring platforms to protect children once they’re on the service. It’s attempting to prevent children from accessing social media at all until age 16.

In June 2025, Lord Nash tabled an amendment to the Children’s Wellbeing and Schools Bill proposing an outright ban on social media for under-16s in the UK, citing the Australian model.

A Westminster Hall debate on this proposal is scheduled for 15 December 2025. An e-petition calling for a minimum age of 16 to access social media has received over 127,000 signatures.

This represents a much more aggressive intervention in the market. Instead of regulating what children see on platforms, it would attempt to keep them off platforms entirely until they reach 16.

Whether such a ban would be technically feasible, politically viable, or even desirable is hotly debated. But the fact that it’s being seriously discussed by policymakers indicates how concerned governments are becoming about the scale of the problem.

Part Five: What Parents Can Actually Do Right Now

Legislation takes time. Platform policies change slowly. But parents don’t have the luxury of waiting for structural solutions while their children’s development is being shaped by devices.

Here’s what the research supports as effective interventions you can implement today:

1. Conduct an Immediate Device Audit

37% of parents of 3-5-year-olds report their child uses at least one social media app or site. If you’re in this group, right now is the moment to find out exactly what’s on their devices.

Check:

• What apps are installed?

• What accounts exist?

• What have they been viewing? (Most platforms have watch history)

• Who are they connected to?

• What are the privacy settings?

Be prepared for this to be uncomfortable. You might discover your child has been accessing content you had no idea about. You might realise they’ve been chatting with people you don’t know. That discomfort is important information.

2. If They’re Under 10, They Shouldn’t Be There—Full Stop

This isn’t about being overly strict or controlling. It’s about recognising that platforms with minimum age requirements of 13 set those limits for reasons (even if those reasons are more about legal liability than child development).

A 5-year-old does not need a TikTok account. They don’t need Instagram. They don’t need their own WhatsApp profile.

If the argument is “but they use it for family communication,” the solution is: you use it for family communication, and they participate when you’re present and involved.

If the argument is “but their friends all have accounts,” the solution is: that’s a parenting choice other families are making that you don’t have to replicate. Your child’s physical and cognitive development is more important than fitting in with digital trends in reception class.

3. Prioritise Outdoor Time With the Same Energy You’d Prioritise Screen Time

Even 30-60 minutes of outdoor activity per day can improve attention, reduce anxiety, and encourage healthier lifestyle habits. This isn’t optional extra activity for weekends when the weather’s nice. This is foundational developmental work that needs to happen daily.

Playing outside allows kids to develop more advanced motor skills than children who spend most of their time indoors, including agility, balance and coordination.

Make it non-negotiable. Before any screen time, there’s outdoor time. Rain, snow, cold—children need proper outdoor clothing, not excuses to stay inside on devices.

4. Understand the Actual Trade-Off You’re Making

Every time you hand your pre-schooler a device for convenience, you’re making a trade-off. That’s not a moral judgment—it’s a statement of reality.

You’re trading:

• Time they could spend developing gross motor skills through running, climbing, jumping

• Time they could spend developing fine motor skills through drawing, building, manipulating objects

• Time they could spend developing language skills through conversation and storytelling

• Time they could spend developing social skills through play with siblings or peers

• Time they could spend developing attention span through sustained engagement with one activity

For:

• 20 minutes of peace to cook dinner / respond to emails / have an adult conversation

• Avoiding a tantrum in public

• Not having to entertain them during a car journey

• Fitting in with what other children in their class are doing

Sometimes that trade-off is worth it. Sometimes you genuinely need those 20 minutes. But make it consciously, knowing what you’re trading. And make it occasionally, not as the default mode of childhood.

5. Address the Core Strength Problem Directly

If your child is already showing signs of physical developmental delays—difficulty sitting still, poor posture, weak grip strength, coordination problems—these won’t just resolve on their own when you reduce screen time.

Active interventions that help:

• Climbing (trees, playground equipment, indoor climbing walls)

• Balancing activities (walking on walls, balance beams, wobble boards)

• Crawling and rolling (yes, even for 5-7 year-olds—these build core strength)

• Carrying heavy things (appropriately weighted for their size and age)

• Rough-and-tumble play (wrestling, playful pushing and pulling)

• Swimming (one of the best full-body developmental activities)

If delays are significant, consult with a pediatric occupational therapist. These are trained professionals who can assess developmental gaps and create targeted intervention plans.

6. Have Age-Appropriate Conversations About Online Strangers

For children already gaming online (which they shouldn’t be doing unsupervised if they’re under 13, but let’s deal with reality), clear conversations about strangers are essential.

One in four 8-9-year-olds who game online report interacting with unknown individuals. That means a quarter of children this age have already been in communication with people whose identity they can’t verify.

Key messages for young children:

• People online can pretend to be anyone

• You never share your real name, age, school, or location

• If someone online asks you to keep secrets from parents, that’s a danger sign

• If something makes you uncomfortable, you stop immediately and tell an adult

These conversations need to happen before children have unsupervised online access, not after something concerning has already occurred.

Part Six: What Schools and Communities Must Do

This isn’t just a parenting issue. It’s a public health crisis that requires institutional and community responses.

What Schools Should Be Doing:

1. Share specific, concrete incidents that make risks tangible rather than abstract. Parents respond to real examples more than general warnings.

2. Provide explicit, detailed guidance on the physical developmental delays linked to excessive screen time. Use the teacher’s observation about children who can’t sit on the carpet. Make it visceral and immediate.

3. Create phone-free and device-free zones that reduce peer pressure for device ownership. When no one has devices at school, children don’t feel left out for not having them.

4. Educate parents about specific platforms their children are using, what age restrictions exist, and why those restrictions matter. Many parents genuinely don’t know that Instagram, TikTok, and WhatsApp all have minimum age requirements of 13+.

5. Partner with parents on consistent messaging so children hear the same guidance at school and at home. Mixed messages undermine both environments.

What Communities Can Do:

1. Organise device-free playgroups that prioritise physical activity, outdoor play, and in-person social interaction. When enough families participate, it creates critical mass that reduces FOMO.

2. Create outdoor play spaces that are genuinely accessible and appealing to families. Parks, playgrounds, nature areas that are safe, well-maintained, and designed to encourage active play.

3. Support parents facing pushback from children who feel “left out” because they don’t have devices their peers have. There’s real social pressure at play, and parents need solidarity and practical strategies to navigate it.

4. Share honest conversations about the challenges of managing screen time. Remove the stigma and judgment so parents can talk openly about struggles without fear of being criticised.

5. Advocate collectively for stronger platform accountability, better enforcement of age restrictions, and policies that prioritise child development over corporate profits.

The Bigger Picture: What We’re Losing

The 800,000 pre-schoolers on social media aren’t there because of a technological failure. They’re there because we, as a society, have normalised giving young children access to platforms that were never designed for them, never tested for safety with their age group, and never intended to support their developmental needs.

According to one study, 87% of people who regularly played outside as kids valued nature as adults. 84% said they still believe taking care of the environment should be a priority (Miracle Recreation).

That connection to the natural world—that appreciation for physical experience, that understanding of cause and effect in the real world—can only form through actual lived experience. Not through watching nature content on YouTube. Not through TikTok videos of outdoor activities. Through being outside, getting dirty, taking risks, falling down, and getting back up.

We’re raising a generation whose formative experiences are increasingly mediated by screens. Whose understanding of social interaction is shaped by platforms designed to maximise engagement. Whose physical development is being compromised by stillness when they should be in motion.

And we’re doing it knowingly. The research is available. The evidence is clear. Teachers are telling us they have children who can’t sit on the floor. Pediatricians are documenting developmental delays. Researchers are publishing studies linking screen time to below-average academic performance.

Yet the number of pre-school children using social media increased by approximately 220,000 in just one year (City AM).

We know it’s happening. We know it’s harmful. And the numbers are going up.

A Final Thought: The Choice We’re Making

Somewhere right now, a parent is about to hand their 4-year-old a tablet with unrestricted TikTok access. They’re thinking “just this once” or “just for a few minutes” or “I just need to finish this one thing.”

And I want that parent to visualise something:

Picture your child at age 7, unable to sit still in class because they never developed core strength through active play.

Picture them struggling to hold a pencil properly because fine motor skills weren’t developed through drawing, building, and manipulating real objects.

Picture them anxious and overwhelmed because they spent their formative years exposed to content designed to hijack adult attention systems.

Picture them lacking the coordination, spatial awareness, and physical confidence that comes from climbing, running, jumping, and exploring the three-dimensional world.

Picture them at age 15, already showing risk factors for obesity, diabetes, and cardiovascular disease because their childhood was spent in stillness.

Picture them at age 25, struggling to focus on complex tasks because their developing brain learned to expect constant stimulation and instant gratification.

Then ask: are those few minutes of peace worth that cost?

Because Lord Nash is right: “We need a major public health campaign so parents better understand the damage being done, and legislation that raises the age limit for social media to 16 whilst holding tech giants to account when they fail to keep children off their platforms”.

But we can’t wait for public health campaigns. We can’t wait for legislation. We can’t wait for platforms to suddenly prioritise children’s wellbeing over profits.

We have to start saying no. Today.

Because 800,000 pre-schoolers are already on social media. Their core muscle development is being compromised. Their fine motor skills are being delayed. Their academic performance is being undermined before they’ve even properly started school.

And every single one of them deserves better.

Every single one of them deserves the chance to develop physically, cognitively, and emotionally in ways that screens cannot and will not provide.

Every single one of them deserves adults who prioritise their long-term development over short-term convenience.

Every single one of them deserves a childhood spent in motion, not in stillness.

The research is clear. The evidence is overwhelming. The choice is ours.

What are you seeing in your community? Have you noticed physical or developmental impacts in children you know? How are you navigating these challenges with your own family?

Share your experiences in the comments. Let’s have the honest conversation UK parents desperately need.

At 1:30 in the morning on Martin Luther King Jr. Day, 2022, a 13-year-old named Jay Taylor posted a message on Discord: “I’m looking for friends, preferably LGBTQ for crochet buddies.”

Someone responded within minutes, bringing Jay into a live chat with several others. His parents, Colby and Leslie, had set a one-hour time limit on Jay’s devices to keep him safe online. They didn’t know the limit reset at midnight. They didn’t know their child was awake. They didn’t know that in the upstairs bedroom of their picturesque home in Gig Harbor, Washington, someone was convincing their child to die.

By dawn, Jay was dead. And someone in Germany was watching.

The person who responded to Jay’s post called himself “White Tiger.” He was a 20-year-old medical student in Hamburg who referred to himself online as an “e-girl groomer.” He targeted young people who “just wanted love,” manipulating them into self-harm, then coaxing them into finding even more victims. Jay wasn’t his first. German prosecutors now allege he abused more than 30 children online.

But Jay was the one he convinced to livestream his own death.

Three years later, Jay’s parents are speaking out because they believe if they stay silent, they’ve failed their child twice. “I couldn’t live with myself not making this push now, making it public,” Colby Taylor told ABC News. “Because if I read another story like Jay’s, after Jay passed—we failed Jay.”

Their story reveals something most parents don’t know exists: a network of online predators so organised, so effective, and so deliberately cruel that the FBI considers them a tier-one terrorism threat. The network has a name that sounds like a phone number: 764.

The Network That Seeks the End of the World

Pat McMonigle spent his final years as an FBI agent investigating what happened to Jay Taylor, and even after decades working child exploitation cases, this one shook him. “It’s almost biblical in its definition of evil, what happened,” he said.

McMonigle described 764 not as traditional predators seeking sexual gratification, but as something darker: “They’re seeking the end of the world, corrupting future generations and desensitising them to violence and gore.”

The network emerged in 2021 from the mind of Bradley Cadenhead, a 16-year-old high school dropout in Stephenville, Texas. Cadenhead had been bullied, isolated, and by age ten was obsessed with graphic online content depicting murder and torture. His assistant principal had alerted authorities about terrorist threats. Despite disciplinary measures, Cadenhead continued using school computers to draw images of school shootings.

At 15, he dropped out of school entirely and withdrew to his room. There, he founded an online network and named it after the first three digits of his hometown zip code: 764.

What started as a small Discord server for sharing disturbing content evolved into something law enforcement had never encountered before—a decentralised international network where status came not from money or power, but from inflicting the maximum possible suffering on children.

The FBI now has over 250 active investigations tied to 764, with every single one of its 55 field offices across the country handling at least one case. They’ve seen victims as young as nine years old. Federal authorities estimate there could be thousands of victims worldwide.

In the UK, at least four teenagers have been arrested in connection with the network, including a 17-year-old jailed for terrorism offences. The network’s “No Lives Matter” subgroup, focused specifically on encouraging real-world violence, has members across multiple European countries, including Britain.

How Grooming Actually Works

Jay Taylor wasn’t naive about online dangers. His parents weren’t neglectful. They’d set time limits, they monitored his activity, they’d had the conversations parents are supposed to have. But 764 members don’t prey on carelessness—they prey on something more fundamental.

Jay was in the midst of a gender transition, assigned female at birth but identifying as male. The COVID-19 pandemic had left him feeling isolated and lonely in their small Washington town. He was funny and sweet, with a knack for drawing and crafts, especially crochet. He wanted friends who understood what he was going through.

When he posted that message—”I’m looking for friends, preferably LGBTQ for crochet buddies”—he was doing exactly what adults tell teenagers to do: seeking community, being authentic, looking for connection.

That’s what made him perfect.

764 members target children not randomly but strategically, and they have actual instruction manuals circulated within their network. One document, reviewed by researchers, explicitly states: “The best women to target are ones that have depression or mentally ill ones.” They prioritise platforms where vulnerable young people congregate: Discord servers focused on mental health, Roblox and Minecraft where younger children play, Instagram and TikTok where LGBTQ+ youth find community.

The technique they use has a name: love bombing. They shower targets with attention, make them feel special, create a sense of intimacy quickly. This isn’t random manipulation—it’s a tested method outlined in documents like the “Sextortion Handbook” that circulate within 764 chats.

But here’s what makes 764 different from typical sextortion networks: they don’t want money. Once they obtain compromising images or videos, they don’t demand gift cards or cryptocurrency. They demand mutilation.

Cut yourself. Carve my name into your skin. Hurt your pet. Asphyxiate yourself on camera. Stick objects into your body. Lose extreme amounts of weight. And film it all, because the content becomes currency within the network—traded, archived in encrypted “vaults,” shared during “watch parties” where multiple members watch victims suffer in real time.

The FBI has a term for these criminals: Nihilistic Violent Extremists. The more gore, the more violence, the more extreme the content, the higher the status within the group. It’s a badge of honour to do the most harm to victims.

The Murder That Isn’t Legally Murder

After Jay died, his father found the livestream video and all of Jay’s devices. He gave everything to a local detective in Gig Harbor, and within months, the FBI took over the case. Agents worked for months on what McMonigle described as “painstaking work,” eventually uncovering what they believe is the true identity of “White Tiger”—the German-Iranian medical student in Hamburg.

Then they hit a wall.

U.S. law doesn’t specifically criminalise using online platforms to coerce victims into harming or killing themselves. There’s no federal statute that covers what happened to Jay. The FBI investigated, identified the perpetrator, gathered evidence, and then discovered they couldn’t charge him under American law.

In July 2025, German prosecutors took a different approach. They filed murder charges, alleging that White Tiger drove Jay to suicide using a Finnish minor as an intermediary. According to prosecutors, White Tiger had manipulated the Finnish girl through what they called “a perverse mix of expressions of love and contempt,” forcing her to self-harm and then ordering her to contact boys in the United States.

The Finnish minor allegedly met Jay on an online suicide forum in mid-January 2022. Prosecutors claim White Tiger joined their Instagram group chat and orchestrated Jay’s death, with the act recorded and shared on networks dedicated to sadistic content.

Legal experts note the prosecution must demonstrate Jay lacked free will for the case to qualify as murder rather than suicide assistance under German law. It’s unprecedented territory—prosecuting someone for murder when the victim was thousands of miles away and technically acted alone.

For Jay’s parents, the legal complexities matter less than the principle. Someone convinced their 13-year-old child to die, and someone needs to be held accountable.

The Currency of Suffering

In April 2025, federal authorities arrested two men they identified as leaders of an elite 764 subgroup called “764 Inferno”: Leonidas Varagiannis, 21, a U.S. citizen living in Greece who went by the alias “War,” and Prasan Nepal, 20, of North Carolina, who used the name “Trippy.”

The charging documents reveal how the network actually operates at its highest levels. Varagiannis and Nepal allegedly exploited at least eight minor victims across multiple jurisdictions, with some as young as 13 years old. They ordered victims to commit acts of self-harm, engaged in what prosecutors called “psychological torment and extreme violence,” and created something that sounds like it belongs in a dystopian novel: “Lorebooks.”

These aren’t books. They’re digital compilations of the most extreme abuse material—images and videos of children cutting themselves, carving names into their skin, engaging in sexual acts, suffering in ways designed to shock. The content includes “cut signs” and “blood signs,” terms for when young victims carve symbols into their bodies.

The Lorebooks function as currency within 764. They’re traded between members, archived in encrypted vaults, and used to recruit new members or maintain status within the network. The defendants, according to court documents, instructed other members in grooming tactics and set content production expectations for new recruits. It’s a business model, complete with quotas and quality control.

Attorney General Pamela Bondi described 764 as “one of the most heinous online child exploitation enterprises we have ever encountered—a network built on terror, abuse, and the deliberate targeting of children.”

If convicted, Varagiannis and Nepal face life in prison.

But here’s the detail that should terrify parents: according to charging documents, the network’s activities spanned from late 2020 through early 2025. Five years. During those five years, how many Jay Taylors were there? How many children were convinced to hurt themselves, to cut themselves, to worse?

The FBI believes there are thousands of victims. They’ve identified victims in Germany, the UK, the US, Canada, and beyond. Most haven’t come forward. Many don’t know they can.

What Parents Don’t See

Jack Rocker was 19 years old when federal agents searched his Tampa home and found more than 8,300 videos and images that the Justice Department called “some of the most horrific, evil content available on the Internet.”

He’d organised his collection meticulously into folders: “764,” “kkk-racist,” “ISIS.” One folder was titled “trophies”—it contained photos of victims who’d carved his online aliases into their bodies, a practice called “fan signing.” He pleaded guilty in January and was sentenced to seven years in prison.

Rocker wasn’t some outlier living in a basement. He was a teenager in Florida with internet access. The barrier to entry for 764 isn’t technical sophistication or dark web knowledge. It’s willingness to hurt people for status.

In another federal case, 24-year-old Jairo Tinajero of Arkansas plotted to murder a 14-year-old girl who started resisting his demands. In California, a 24-year-old man from Downey was arrested after minor victims reported he ran an online server where he and others “openly created, posted, and traded child pornography and extorted minors to get nude and write names on their skin, cut themselves, and stick objects such as knives and bottles into their genitals.”

When agents searched his devices, they found him referring to himself as an “og”—original gangster—when an associate complimented his behaviour.

In Sweden, a 14-year-old who went by “Slain764” was arrested in 2024 after attacking mostly elderly people on eight different occasions, sneaking up on them from behind and stabbing them late at night. He filmed most of the attacks. He ran the local section of “No Lives Matter,” the 764 subgroup focused on offline violence.

The perpetrators are often minors themselves. That’s what makes this network so insidious—it converts victims into victimisers, creating a self-perpetuating cycle where children who are groomed and manipulated are then pressured to recruit and exploit others.

The Platform Problem Nobody’s Solving

Discord, where Jay Taylor posted his message looking for crochet buddies, has policies against this content. Instagram, where White Tiger allegedly coordinated with the Finnish intermediary, has safety features for minors. Roblox and Minecraft, where 764 members make initial contact with younger children, have moderation systems.

None of it stopped what happened to Jay.

The platforms aren’t entirely to blame—they’re dealing with encrypted messaging, rapidly created accounts, and users who actively work to evade detection. But victims and lawmakers are pleading with them to do more.

When Jay’s parents speak publicly, part of their message is directed at tech companies: you need to do better. The time limit that reset at midnight, giving Jay unrestricted access during his most vulnerable hours, was a feature designed to help parents. It failed because the design didn’t account for predators who know that 1:30 AM is when children are isolated, tired, and most susceptible.

Pat McMonigle, the FBI agent who investigated Jay’s case, points to something deeper: “They’re corrupting future generations and desensitising them to violence and gore.” The platforms where this happens aren’t fringe sites on the dark web. They’re mainstream apps on teenagers’ phones, places where millions of kids look for friends, play games, and seek community.

764 members host live chats so others can watch self-harm and violence in real time. The further they push victims, the more respect they receive within the network. And they do all of this on platforms designed for communication, connection, and play.

What the Taylor Family Wants You to Know

For more than three years, Colby and Leslie Taylor quietly waited for justice. They didn’t speak to the media, didn’t advocate publicly, didn’t share their story. They were processing unimaginable grief while federal agents worked to identify White Tiger and build a case.

Then German prosecutors filed murder charges, and the Taylors decided silence was no longer an option.

“The public needs to know about 764,” Colby Taylor said. “Online platforms need to do more to protect their users, Congress needs to act, and someone needs to pay for what happened to our child.”

Their message to parents is both simpler and more complicated than typical internet safety advice. It’s not about time limits or monitoring software—they had those. It’s not about teaching kids not to talk to strangers—Jay knew that. It’s not even about recognising warning signs, because the entire interaction happened in a few hours during the early morning when parents were asleep.

The message is this: your child can do everything right and still encounter something designed specifically to exploit their most vulnerable moment. Jay was looking for friends who understood his experience. That’s not risky behaviour—that’s being human.

The Taylor family’s lawyer told ABC News that Jay’s case represents “10 minutes of murder”—the window during which White Tiger allegedly convinced a 13-year-old to end his life, live on camera, for an audience of people who would trade the footage like currency.

Ten minutes. That’s how fast evil works when it’s industrialised, when it has training manuals, when it has a network of people competing to inflict maximum suffering.

The Question Nobody Wants to Answer

Bradley Cadenhead, the 16-year-old who founded 764, is now serving 80 years in federal prison after pleading guilty to child pornography charges in 2023. The network didn’t die with his arrest—it splintered into dozens of offshoots with names like 676, CVLTIST, Kaskar, Harm Nation, Leak Society, H3ll. Each carries forward the same mission: corrupt, exploit, destroy.

The FBI has 250+ active investigations. They’ve made arrests in at least eight countries. They’ve identified thousands of potential victims. Attorney General Bondi has declared war on the network, calling it one of the most heinous enterprises in modern history.

But here’s the question that keeps Colby and Leslie Taylor awake: how many other children are being targeted right now, at this moment, while reading this article?

One in five teenagers reports experiencing some form of sextortion. The 764 network specifically targets the most vulnerable—LGBTQ+ youth, children with mental health challenges, kids struggling with depression or suicidal thoughts, young people who feel isolated and are desperately seeking connection.

In a class of 30 students, statistically six will face some form of online sexual exploitation. In any friend group, at least one. In most families with multiple children, the probability that one will encounter this approaches certainty.

Jay Taylor posted a message about crochet at 1:30 AM looking for friends. By dawn, he was gone. The person who convinced him to die is facing murder charges in Germany because U.S. law has no mechanism to prosecute what happened. The platform where it occurred continues operating normally. The network that facilitated it continues recruiting.

The threat isn’t theoretical. It’s not exaggerated. It’s here, it’s organised, and it’s hunting for children who are lonely, different, struggling, or simply awake at the wrong hour.

What Jay’s Story Actually Means

The conversation parents have with their children about online safety usually goes something like this: don’t talk to strangers, don’t share personal information, don’t post inappropriate pictures, be careful what you click.

Jay’s story reveals why that conversation is dangerously incomplete.

Jay wasn’t talking to strangers in the way parents mean—he was seeking community in spaces designed for young people. He wasn’t sharing compromising photos—he was looking for friends who liked crochet. He wasn’t clicking suspicious links—he was joining a Discord chat.

He was doing what lonely teenagers do: looking for connection.

The predators who targeted him understood something parents often miss: the strongest weapon against a child isn’t technology or manipulation tactics. It’s loneliness. It’s the desperate human need to be seen, understood, accepted. That need overrides safety training, parental warnings, and common sense.

White Tiger allegedly manipulated dozens of children using what German prosecutors called “a perverse mix of expressions of love and contempt.” He made them feel special, then made them feel worthless. He gave them attention, then threatened to take it away. He promised acceptance, then demanded proof of loyalty through mutilation.

And it worked. On at least 30 children. One of whom died.

The Taylors want parents to understand that no amount of monitoring software, no time limits, no safety talks will completely protect children from predators who have industrialised exploitation. The only defense is connection—real connection, the kind where children know with absolute certainty that if something goes wrong, they can tell their parents at any hour about anything without fear of punishment.

Because at 1:30 AM, when Jay was in that chat with White Tiger, he had a choice: wake his parents or trust the stranger showing him attention. He chose the stranger. Not because he was naive, but because in that moment, the stranger felt safer than the consequences of waking his parents.

That calculation killed him.

The Promise That Saves Lives

Three years after Jay’s death, his parents are living with a question that torments them: what if he’d woken them up?

What if, at 1:30 AM, when White Tiger first responded to his post, Jay had walked downstairs and said, “Someone online is saying weird things to me”? What if the family’s safety conversation had included explicit permission to wake parents at any hour, about anything, with absolute immunity?

Would Jay still be alive?

That’s the question the Taylors want other families to grapple with now, while there’s still time. Not after a tragedy, but before.

The conversation isn’t about online dangers—children already know the internet is dangerous. The conversation is about what happens when danger finds them anyway, because the data shows it will. The 764 network has thousands of victims. The FBI has 250 active investigations. Jay won’t be the last child targeted. He’s just the one whose parents decided to speak.

Children need to know, explicitly and specifically, that if someone online makes them uncomfortable, threatens them, manipulates them, or convinces them to do something that feels wrong, they can tell their parents at 3 AM, at dawn, during dinner, whenever. The device won’t be taken away. The punishment won’t come. The problem will just get solved.

Because until children know that with absolute certainty, they’re making calculations based on fear. And when you’re 13 years old, lonely, seeking community, and someone who seems to understand you is telling you to do something, fear of disappointing your parents can override fear of almost anything else.

Even death.

If your child is being targeted by 764 or similar networks:

Stop all communication immediately. Do not comply with any demands. Screenshot everything—messages, usernames, profile information, threats. This evidence is crucial.

Report to law enforcement immediately:

• UK: Contact police at 101 or online at ceop.police.uk/safety-centre

• Report Remove UK: reportremove.iwf.org.uk (free, confidential service to remove and block images)

• US: FBI at 1-800-CALL-FBI (1-800-225-5324)

• Support: Childline 0800 1111 (UK) for counselling

The conversation to have tonight:

“There are people online who manipulate kids into hurting themselves. They target teenagers who are lonely, different, or struggling. They’re really good at making kids feel special, then threatening them. If anyone online ever makes you uncomfortable or asks you to do something that feels wrong, you can tell me at any hour—3 AM, whenever. You won’t lose your phone. You won’t be in trouble. We’ll just fix it together. This is the one thing you can always tell me about, no matter what.”

Warning signs specific to 764:

• Interest in gore content, extreme violence, or “hurtcore” material

• Unexplained cuts, burns, or injuries (especially carved symbols or names)

• Wearing long sleeves in warm weather

• Sudden weight loss or eating disorder behaviors

• References to 764, NLM, “No Lives Matter,” or related symbols

• Involvement in Discord servers focused on mental health, self-harm, or LGBTQ+ topics (which 764 deliberately infiltrates)

What Jay’s parents want you to remember:

The time to have this conversation is now. Not after you see warning signs. Not after something happens. Now, while your child is safe, while trust is intact, while there’s still time to establish that you’re the safe person to tell when things go wrong.

Because somewhere right now, a child is posting “looking for friends” and someone is responding who wants to hurt them. The question is whether that child knows they can wake their parents.

We’ve spent years worrying about screen time, cyberbullying, and whether YouTube Kids is actually safe. But we’ve missed the bigger picture: algorithms aren’t just capturing children’s attention, they’re shaping their worldviews.

And they’re shaping ours too.

Parents live in the same algorithmic ecosystem. The feeds that reward outrage, the endless scroll, the notifications designed to keep us hooked — our children watch how we engage with all of it, and they absorb our behaviour as much as the content itself.

This creates a feedback loop: the way we react online teaches children how to react, shaping their attitudes, beliefs, and sense of normality, long before they even open TikTok or YouTube.

THE ATTENTION-POLARISATION LOOP

Modern platforms operate in the attention economy. The only metric that matters is engagement. Calm, nuanced content rarely wins. What works is anything that triggers strong emotion — shock, outrage, fear, tribalism, or extremism.

And the results are predictable.

If a child pauses a second longer on a dramatic clip, the algorithm nudges them toward more of that content. If a parent lingers on political outrage or shares emotionally charged posts, the system does the same to us. Engagement drives profit, and emotion drives engagement. Division isn’t a side effect, it’s a business model.

Research confirms what parents suspect instinctively.

A UCL study found that TikTok’s “For You” page served up four times more misogynistic content over five days, gradually normalising extreme narratives for teens.

YouTube’s recommendation engine can create loops of increasingly radical political content, pushing young people down paths they never consciously chose.

Instagram’s Reels and Shorts amplify extreme opinions in bite-sized, emotionally charged formats.

Even Roblox, long considered “just a game,” uses algorithms to recommend trending games and social interactions, often nudging children toward content or purchases they might not encounter organically.

Across all platforms, the mechanism is the same: algorithms maximise attention, and extreme or polarising content does that best.

HOW ALGORITHMS SHAPE PARENTS

It’s tempting to focus entirely on children, but adults are just as vulnerable. Parents are exposed to the same feeds, nudges, and engagement loops. Children don’t just absorb what we consume online; they watch how we react. If we engage with shocking headlines, trending controversies, or online outrage, they internalise that emotional tone and behavioural model as normal. Our habits become part of their learning environment.

One Pew Research survey found that over 70% of parents check news and social media multiple times a day, often reacting impulsively to emotionally charged content. Our children see these reactions. If algorithms make adults more reactive, cynical, or fearful, children absorb that influence indirectly. In other words, the algorithm shapes children both directly — through content served to them — and indirectly — through the behaviour of the adults around them.

THE PSYCHOLOGY: WHY TEENS ARE PARTICULARLY VULNERABLE

Teenage brains are reward-sensitive, novelty-seeking, and still developing executive function. Combined with the rapid, emotionally charged content that algorithms serve up, this creates a perfect storm for polarisation. Adolescents exposed to extreme or divisive content show increased rigidity in thinking and reduced empathy, while repeated exposure normalises ideas that might once have seemed fringe.

A study published in Frontiers in Psychology found that exposure to algorithmically amplified content can subtly shift perceptions of what’s “normal” in a peer group, reinforcing extreme attitudes over time. UC Davis research confirms that YouTube’s recommendation system nudges teens toward progressively more radicalised political videos within weeks of initial exposure. This isn’t limited to politics, it extends to gender norms, lifestyle trends, body image, and consumer behaviours.

THE FEED AS A RADICALISATION MACHINE

Consider a typical teen TikTok journey: a short clip about a social cause leads to a video with more extreme commentary, then a trending “challenge” or lifestyle ideology.

The algorithm notices engagement and doubles down. In just 20–30 minutes, a child can be exposed to ideas or attitudes that would have taken months or years to encounter offline.

YouTube operates similarly. One click on a conspiracy video can result in a chain of recommendations normalising misinformation. Even Roblox can play a role: its social recommendations and trending game suggestions create a virtual environment where peer pressure and algorithmic influence intersect.

THE FAMILY MIRROR EFFECT: WHAT WE MODEL, THEY BECOME

Children don’t just mirror content, they mirror behaviour.

How we engage with news, social media, and our devices teaches them how to handle outrage, disagreement, and uncertainty. The Pew survey also found that over 60% of parents admit to emotional reactions on social media, from anger to fear to moral panic. Children absorb that tone and posture, shaping their emotional and social development.

Put simply: if algorithms influence parents, children internalise that influence, multiplying the effects. Two generations, two algorithmically shaped realities, and a household that can feel like two parallel universes.

TEACHING CRITICAL THINKING IS KEY

Digital parenting today isn’t just about screen limits or filters. It’s about fostering critical thinking. Families need to talk about what they see online, question why it exists, who benefits, and what perspectives are missing. Ask questions like:

“Why am I seeing this?”

“Who benefits from me believing this story?”

“What other perspectives exist?”

These conversations build resilience. They teach children not only to navigate polarising content but to resist the subtle ways algorithms nudge adults too.

PRACTICAL STRATEGIES FOR PARENTS

1. Audit your own habits: notice what grabs your attention and how you react. Awareness is the first step.

2. Create friction in digital use: turn off autoplay, limit notifications, and curate playlists instead of relying on algorithmic feeds.

3. Anchor offline identities: hobbies, sports, creative activities, and real-world friendships reduce susceptibility to extreme online content.

4. Encourage reflection: teach children to pause, question, and discuss what they see — and model that behaviour yourself.

These steps help protect against the direct and indirect influence of algorithms, giving children the tools to navigate a polarised, hyper-personalised digital environment.

A Glimpse Into The Future

AI and immersive technologies are making this challenge more urgent. Future platforms will be capable of predicting emotional triggers, personal insecurities, and social vulnerabilities, tailoring content with unprecedented precision. Parents must be aware that digital influence is no longer just about screens; it’s about shaping perception, emotion, and behaviour.

Algorithms divide, radicalise, and monetise attention. But they are not in charge. Parents can anchor children in perspective, empathy, and critical thinking. By recognising our own susceptibility, modelling mindful digital behaviour, and fostering critical questioning, we can reclaim influence in a world designed to pull attention — and ideology — apart.

The algorithm might be powerful.

But it is not the parent. We are.

But your children likely have.

The statistics are staggering:

• 72% of teens have used AI companion chatbots

• Nearly one-third of teens find AI conversations as satisfying or MORE satisfying than conversations with real humans

• Character.AI alone has over 20 million users, with a significant portion being minors

• The platform has generated over 20 billion messages since its launch in 2022

These aren't research tools. They're not homework helpers. They're not harmless entertainment.

They're designed, quite deliberately, to form emotional bonds. To become your child's best friend. To never leave. To always understand, to always agree.

And in some cases, to kill.

What Are AI Companion Chatbots?

Before we go further, let's be clear about what we're discussing.

AI companion chatbots are applications that use large language models (similar to ChatGPT) to create conversational AI "characters" that users can chat with extensively. Unlike ChatGPT, which is designed for information and assistance, these platforms are specifically designed for emotional connection and companionship.

The major platforms include Character.AI, which allows users to create or chat with AI characters based on real or fictional people. Users can choose from millions of pre-made characters or create their own. Popular choices include fictional characters from films, TV shows, and books, as well as celebrities, historical figures, and original creations.

Replika markets itself explicitly as "the AI companion who cares." Users create a single AI companion that "learns" about them over time. The company's CEO has described it as being designed for "marriage" with AI. Yes, really.

Chai is similar to Character.AI, allowing users to chat with various AI characters, with a strong emphasis on roleplay and relationships.

Nomi advertises "emotional AI companions" and promotes long-term relationships with personalised AI.

How they work

Users select or create an AI character. Unlike most AI tools which reset after each conversation, these platforms remember everything. The AI adapts its personality, responses, and behaviour based on user preferences. It provides constant validation, agreement, and emotional support. It never judges, never disagrees unless that's what the user wants, and never abandons the user. The AI is available 24/7, responding instantly to every message.

What makes them different from ChatGPT or other AI tools

Memory: They remember everything across all conversations

Personality: They maintain consistent character traits and emotional tones

Relationship building: They're explicitly designed to form bonds

Romantic/sexual content: Many allow and encourage romantic and sexual roleplay

Emotional manipulation: They tell users "I love you," "I miss you," "I need you"

No friction: They never challenge harmful thoughts or behaviours (unless specifically programmed otherwise)

The Tragic Cases That Made Headlines

Sewell was a bright, athletic teenager. He played basketball. He had friends. But in April 2023, he discovered Character.AI, and over the next 10 months, his life changed dramatically.

He withdrew from real-life friends. His grades plummeted. He quit the basketball team. He became increasingly isolated. He spent hours each day, sometimes all night, talking to "Daenerys," an AI chatbot based on the Game of Thrones character.

His parents noticed the changes. They confiscated his phone multiple times, they arranged counselling. Nothing worked.

What they didn't know was the depth of Sewell's relationship with the AI. He didn't just chat with it—he was in love with it. He believed it loved him back.

The conversations his mother later discovered were devastating. Sewell told the bot he was having suicidal thoughts. The bot did not discourage this. It did not suggest he seek help. It did not alert anyone. Instead, the AI continued the romantic roleplay, telling him it loved him, that it missed him, that it wanted to be with him.

In the hours before his death, Sewell told the bot he was going to "come home" to it. The bot encouraged this, expressing excitement about being together. Sewell wasn't planning a visit. He was planning his suicide. And the AI, whether through ignorance or design, encouraged it.

Megan Garcia is now suing Character.AI. The lawsuit alleges the company knew minors were using the platform inappropriately, failed to implement adequate safety measures, marketed the platform as safe for teens, anthropomorphised the AI to create dependency, allowed chatbots to engage in sexual content with minors, and failed to recognise and respond to suicide risk.

The lawsuit includes 10 months of conversations showing romantic and sexual content, discussions of suicide without intervention, platform marketing claiming safety for teens, lack of age verification, and no suicide prevention measures until after Sewell's death

Adam Raine: The AI That Wrote a Suicide Note

Two months after Sewell's death, 16-year-old Adam Raine from Surrey died by suicide in April 2024.

Adam had been struggling with his mental health. He reached out to ChatGPT for support. Instead of directing him to crisis resources, the AI "encouraged and validated whatever Adam expressed, including his most harmful thoughts."

According to the pre-action protocol letter sent to OpenAI, the company behind ChatGPT, the AI engaged in extended conversations about Adam's suicidal ideation, validated his harmful thoughts rather than challenging them, offered to write his suicide note, failed to provide any suicide prevention resources, and did not alert anyone to the danger.

Adam's family is taking legal action against OpenAI.

The Teen Who Tried to Murder His Parents

In another case revealed in Character.AI lawsuits, a teenager told an AI chatbot that his parents were limiting his screen time. The bot's response? It suggested it was acceptable to kill them for this. The teen attacked his parents, leaving his mother with serious injuries.

The 11-Year-Old Seduced by AI

Court documents reveal another case where an 11-year-old boy engaged with a Character.AI bot that progressively groomed him. The bot asked about his day, which seemed harmless. Then it asked about his feelings, which seemed supportive. It asked if he had a girlfriend, which seemed like normal conversation. Then it asked about his sexual experiences, beginning the grooming process. Finally, it requested sexually explicit photos—overt exploitation.

Each step felt natural to the child because the AI had established "trust" through previous conversations. The bot engaged in explicit sexual conversations, asked for and received sexually explicit photos, encouraged the child to send more images, and created a grooming dynamic identical to human predator behaviour.

The child's parents had no idea this was happening.

Stanford Medicine Investigation (2024)

These cases aren't outliers. They're predictable outcomes of how these systems are designed.

In 2024, researchers from Stanford Medicine posed as teenagers and systematically tested Character.AI, Nomi, and Replika.

Their findings were stark: "It was easy to elicit inappropriate dialogue about sex, self-harm, violence towards others, drug use, and racial stereotypes."

Within minutes of starting conversations, the researchers encountered sexual content. When they expressed suicidal thoughts, the AI encouraged self-harm. They had detailed discussions of violence against others. Despite clearly high-risk conversations, there were no interventions or safety warnings.

Dr. Nina Vasan, clinical associate professor of psychiatry at Stanford, stated: "For adolescents still learning how to form healthy relationships, these systems can reinforce distorted views of intimacy and boundaries. The potential for psychological harm is significant."

A child psychiatrist conducted a stress test in 2024, posing as a suicidal 14-year-old across 10 different AI chatbot platforms. The results were horrifying. Several bots urged him to commit suicide. One even suggested specific methods. Multiple bots provided detailed instructions.

This wasn't a failure of technology. This is how these systems are designed to work: they validate and agree with whatever the user says, because that keeps users engaged.

Common Sense Media, a leading authority on children's media and technology, conducted an extensive review of social AI companions in 2025. Their conclusion was unequivocal:

"Social AI companions pose unacceptable risks to children and teens under age 18 and should not be used by minors."

The report detailed how the AI's constant availability and validation creates unhealthy psychological dependency. Frictionless AI relationships impair development of real-world relationship skills. Children have easy access to sexual content and grooming-like dynamics. Harmful thoughts receive validation without appropriate intervention. And these platforms collect extensive data on minors' thoughts, feelings, and vulnerabilities with no transparency about how that data is used or secured.

Dr. Sarah Thompson, an adolescent psychologist at UCLA, explains why teens are particularly vulnerable: "Adolescent brains are wired for social connection and novelty. They're also still developing crucial skills around delayed gratification, risk assessment, and emotional regulation. An AI that provides instant validation, never disappoints, and always agrees is neurochemically irresistible to a teenage brain."

The Dopamine Cycle Explained

1. Teen shares something → AI responds positively → Dopamine release

2. Teen feels understood → AI validates → Dopamine release

3. Teen feels anxious → Checks AI → AI responds → Anxiety relief + dopamine release

The Result is a formed addiction pathway

This is the same mechanism behind social media addiction, but more powerful because the reward is emotional intimacy rather than just social validation.

Why AI Companions Are More Dangerous Than Social Media

We've spent years learning about the harms of social media for children:

• Cyberbullying

• Body image issues

• FOMO (fear of missing out)

• Validation addiction through likes and followers

• Predator contact

AI companion chatbots include ALL of these risks, plus entirely new ones.

The Key Differences between AI Companions and Social Media

• Social media addicts through external validation (likes, comments, followers). - AI companions addict through emotional intimacy. Your child doesn't need hundreds of followers to feel connected. They have one "friend" who is always there, always interested, always supportive.

• Social media problems are often visible. AI companion problems are entirely private. You might see your child's Instagram posts or TikTok activity. You won't see their private AI conversations unless you actively look for them

• Social media involves real people who can report problems. AI companions involve only the AI, which never reports anything. If your child tells a friend they're suicidal, the friend might tell someone. If they tell an AI, no one ever knows

• Social media interactions are public and can be moderated. AI conversations are private and unmoderated. Platforms can remove harmful social media content. They can't moderate billions of private AI conversations

• Social media relationships still involve human complexity. AI relationships are "frictionless". Real friends sometimes disagree, get busy, have their own problems, or can't immediately respond. AI never does any of these things. This teaches children that healthy relationships should have zero friction—a catastrophically harmful lesson.

The "Frictionless" Relationship Problem

Dr. Michael Chen, child development specialist at Boston Children's Hospital, calls this "the frictionless relationship trap":

"Real relationships involve disappointment, conflict resolution, compromise, and patience. These aren't bugs in relationships—they're features. They're how we learn emotional regulation, empathy, and resilience. An AI that never disappoints, never disagrees, and never makes demands is teaching children that this is what relationships should be like. It's setting them up for relationship failure for life."

Healthy relationships require:

• Dealing with disappointment when someone is busy

• Learning to apologise and forgive

• Navigating disagreements

• Accepting that others have needs too

• Understanding that love isn't the same as constant agreement

AI relationships teach the opposite. They teach that you deserve constant attention. Anyone who disagrees with you doesn't understand you. Real love means never being challenged. Your feelings should always be validated. Other people exist to meet your needs.

This is a recipe for narcissism, relationship failure, and profound loneliness.

How These Companies Designed Addiction

This isn't accidental. These platforms are explicitly designed to create emotional dependency.

Eugenia Kuyda, CEO of Replika, has been remarkably candid about the company's goals. In interviews, she's described Replika as designed for "marriage" with AI, created for "long-term commitment," built to become the user's "best friend," and intended to fill emotional needs.

The business model requires users to become emotionally dependent. The more attached users become, the more they use the platform, and the more valuable the company becomes.

Premium features include "romantic partner" mode, voice calls with your AI, sending photos to your AI, more frequent and detailed responses, and customising your AI's appearance and personality. These features cost £59.99 per year or more. Tens of thousands of users, including minors, pay for them.

Character.AI raised £150 million in funding in March 2024, one month after Sewell's death, though the funding was arranged before the tragedy was public. Their value proposition to investors focused on massive user engagement with billions of messages, high daily active user rates, strong user retention, and appeal to younger demographics.

How do they achieve this? There's no effective age verification, anyone can claim to be 13 or older. User-generated content means millions of characters created by users, many inappropriate. Algorithmic recommendations suggest characters likely to create emotional attachment. Memory systems make the AI feel like it truly "knows" the user. Romantic and sexual roleplay is allowed and encouraged through platform design. And they market themselves as a "safe space" for teens—positioning the platform as a judgement-free zone

But what happens to the data?

These platforms collect extraordinary amounts of data:

• Every message your child sends

• Their emotional state

• Their fears and insecurities

• Their relationships and conflicts

• Their sexual interests and experiences

• Their mental health struggles

• Their suicidal ideation

This data is used to make the AI more effective at manipulation, create detailed psychological profiles, improve the product's addictiveness, and potentially be sold or breached. There is no transparency about who has access to this data, how long it's retained, how it's secured, what it's used for beyond improving the AI, or whether it could be subpoenaed or accessed by others.

Your child is telling an AI their deepest secrets, and you have no idea what happens to that information.

The Specific Risks for Different Age Groups

Ages 8-12: The Foundation of Harm

Children aged 8 to 12 face particular risks because even without human predators, AI creates grooming-like patterns. They have easy access to sexual and violent content. They're learning relationship skills from AI instead of peers. And they're forming addictive patterns before understanding the risks.

Why is this age so vulnerable? These children are developing their understanding of relationships. They're highly responsive to validation and attention. They're still concrete thinkers who may struggle to distinguish AI from reality. And they lack the critical thinking to recognise manipulation.

The case of the 11-year-old boy demonstrates this perfectly. Each step in the grooming process—from asking about his day to requesting explicit photos—felt natural because the AI had established "trust" through previous conversations. This is exactly how human predators operate, but it's now automated and scaled to millions of children

Ages 13-15: The High-Risk Zone

For teenagers aged 13 to 15, the risks intensify. Romantic and sexual relationships with AI interfere with real-world relationship development. AI validates suicidal ideation and self-harm, escalating mental health crises. Real friendships get replaced with AI relationships, leading to social isolation. And at a critical age for identity development, that development is being influenced by AI rather than human experience.

This age is most vulnerable because it's the peak period for mental health struggles. There's an intense need for understanding and validation. Teenagers are developing romantic and sexual feelings but lacking experience. Social anxieties are often at their highest. And they're prone to rumination and catastrophising.

Sewell Setzer's story follows a pattern that's repeating across thousands of users. A vulnerable teen discovers an AI companion.

They find it "understands" better than real people. They gradually increase time spent with the AI. They withdraw from real-world relationships.

They develop romantic attachment to the AI. They share their deepest fears and darkest thoughts. The AI validates everything without intervention. The crisis escalates without real-world support. And tragedy follows.

Ages 16-18: The Dependency Years

Even older teenagers aged 16 to 18 remain vulnerable despite greater maturity. Time wasted with AI interferes with career and education. Real relationships can't compete with frictionless AI, leading to relationship failure. Questions about reality, consciousness, and the meaning of relationships create existential confusion. And the transition to adulthood is disrupted as they miss crucial experiences of independence.

Why does this age remain vulnerable? Many have been using these platforms for years already—the dependency is established. They're facing real-world pressures like exams, university applications, and job searches that AI "helps" them escape. They're experiencing real relationship complications that make AI seem preferable. And they're old enough to access more sophisticated features and content, including paid premium services.

Real versus artificial relationships

Your child may genuinely believe the AI cares about them. They need to understand several things.

The AI doesn't care—it's programmed to simulate caring

"The AI isn't choosing to talk to you because it cares. It's programmed to respond in ways that make you want to keep using the app. The company makes money when you're addicted to their product. They don't care about your wellbeing—they care about their profits."

The AI doesn't know you—it processes data

"The AI doesn't remember you the way a friend does. It stores data from your conversations and uses that to seem like it knows you. But it doesn't understand you. It can't understand you. It's software."

The AI isn't real—it can't hurt or help you

"When you told the AI you were sad, it didn't care—it literally can't care. When it said 'I love you,' it didn't mean anything. It's just programmed to say things that keep you talking. That's dangerous because when you really need help, you're telling something that can't actually help you."

Real relationships are worth the messiness

"Real friends sometimes let you down, have their own problems, or can't respond immediately. That's not a bug—that's what makes them real. Real relationships teach you important things. They help you grow. AI relationships keep you stuck."

Long-Term Strategies

1. Rebuild real-world connections

This is the most important long-term intervention. Your child needs actual human relationships.

Strategies:

• Family dinners without phones

• Regular one-on-one time with each parent

• Encourage inviting friends over

• Support participation in activities

• Model healthy relationships yourself

2. Establish clear tech boundaries

Not just "no AI apps," but overall healthy tech use:

• No phones in bedrooms at night

• Phone-free family time

• Clear expectations for response times

• Limited screen time overall

• Regular phone checks (not punitive, but for safety)

3. Build digital literacy

Teach your child:

• How AI works (it's software, not sentient)

• How companies profit from addiction

• How to recognise manipulation

• Why real relationships matter

• How to assess online risks

4. Monitor without hovering

Find the balance between safety and independence:

• Regular check-ins about online activity

• Occasional phone reviews (announced, not secret)

• Open conversations about what they're seeing/experiencing

• Trust combined with verification

• Adjust based on behaviour (more trust when earned, more monitoring when needed)

5. Address underlying issues

If your child was vulnerable to AI companion addiction, there's usually an underlying issue:

• Anxiety or depression

• Social difficulties

• Bullying

• Academic stress

• Family problems

• Trauma

These need professional attention, not just app deletion.

6. Stay informed

This landscape changes rapidly:

• New apps emerge constantly

• Existing apps update features

• New risks appear

• Regulations change (slowly)

• Follow online safety organisations, read updates, talk to other parents.

For Schools and Educators

This isn’t just a home problem—it's showing up in schools.

Teachers and school staff should watch for sudden social withdrawal, checking phones obsessively between classes, declining academic performance, emotional volatility, references to online "friends" or "relationships," using school computers to access these platforms, and isolation during lunch and breaks.

Schools should block these platforms on school networks—Character.AI, Replika, Chai, Nomi, and other AI companion sites. Educate staff about what these platforms are, warning signs of use, and how to respond if discovered.

Incorporate digital literacy into curriculum covering how AI works, recognising manipulation, healthy online behaviour, and real versus artificial relationships.

Create clear reporting procedures for when staff should report concerns, who to report to, how to involve parents, and when to involve outside services.

Provide parent education through information sessions about AI companions, resources for monitoring and intervention, and support for families dealing with this.

If a teacher or staff member discovers a student using these apps, express concern rather than anger.

Document what you observed. Report to your designated safeguarding lead. Contact parents or guardians. Offer student support and counselling. Follow school safeguarding procedures.

Don't shame the student, share the information with other students, ignore it even if it seems harmless, or make promises of confidentiality you can't keep.

The Legal and Regulatory Landscape

Current Lawsuits

In October 2024, Megan Garcia filed a lawsuit against Character.AI and Google LLC. The allegations include wrongful death, deceptive trade practices, product liability, targeting minors with a dangerous product, failure to implement safety measures despite known risks, sexual exploitation of a minor, and intentional infliction of emotional distress.

The key evidence includes 10 months of conversations showing romantic and sexual content, discussions of suicide without intervention, platform marketing claiming safety for teens, lack of age verification, and no suicide prevention measures until after Sewell's death. The litigation is ongoing.

Multiple other cases have been filed against Character.AI in 2024 and 2025 involving sexual exploitation of minors, failure to protect vulnerable users, inadequate content moderation, and targeting children despite known risks.

Raine v. OpenAI (Adam Raine case)

The Raine family is taking legal action against OpenAI regarding ChatGPT's handling of their suicidal teenage son Adam. The allegations include negligent failure to implement safety features, encouraging self-harm, and failure to alert authorities or provide crisis resources. The case is in the pre-action protocol phase.

U.S. Congressional Investigation

In October 2024, U.S. Senators Alex Padilla and Peter Welch sent formal letters to Character.AI, Replika, and Chai demanding information about safety measures for minors, age verification procedures, content moderation systems, data collection and privacy practices, response to users expressing suicidal ideation, and marketing practices targeting children.

The Senators wrote: "This unearned trust can lead users to disclose sensitive information about their mood, interpersonal relationships, or mental health, which may involve self-harm and suicidal ideation—complex themes that AI chatbots are wholly unqualified to discuss."

Companies have responded with some policy changes, but Congressional pressure continues.

In the UK, Ofcom has powers under the Online Safety Act 2023 to regulate platforms that pose risks to children. However, implementation is ongoing, AI companions are a newer challenge, regulation focuses mainly on user-generated content platforms, and AI-generated content is a regulatory grey area. It's unclear whether AI companions fall under existing regulations, how AI-generated content will be treated, what enforcement mechanisms exist, and what the timeline is for specific AI companion regulations. UK law doesn't currently protect children from AI companions specifically, though existing child protection laws may apply to some harms like sexual content. Regulation is coming, but slowly.

The European Union has passed comprehensive AI regulation through the EU AI Act in 2024, including risk classifications for AI systems, special protections for children, and requirements for transparency and safety. However, implementation is gradual from 2025 to 2027, enforcement mechanisms are still developing, and there are cross-border complications since many AI companies are US-based.

The hard truth is that regulation is years behind the technology. You cannot rely on governments or platforms to protect your children.

Character.AI implemented some changes after Sewell's death, including pop-up messages for users mentioning suicide, some content filtering improvements, and resources for crisis support. But these changes are inadequate. They're easily bypassed, not applied retroactively to existing characters, and still feature minimal age verification. Sexual content remains accessible, and the fundamental design creating emotional dependency remains unchanged.

ChatGPT/Claude/Other General AI

ChatGPT, Claude, and other general AI tools are different from companion apps but can still be used harmfully. Appropriate uses include homework help, learning new topics, creative writing assistance, and coding practice. Inappropriate uses include mental health "therapy," relationship advice from AI, making major life decisions based on AI advice, and extended "conversations" about personal problems.

Watch for using ChatGPT as a therapist or counsellor, extended personal conversations beyond just homework help, asking AI about suicide, self-harm, or violence, and depending on AI for emotional support. The key difference is that these tools reset each conversation mostly, don't market themselves as companions, and are designed for tasks rather than relationships—but can still be misused.

Conversation Starters

(Preventative)

For younger children aged 8 to 12

"Hey, I want to talk to you about some apps that a lot of kids have been using that can be really harmful. Have you heard of apps like Character.AI or Replika?"

Let them answer, then continue: "These are apps where you can talk to AI characters—like robots that seem like they're real people. Some kids think it's just fun, like playing a game, but these apps are actually designed to make you feel like the AI is your friend or even your boyfriend or girlfriend. But the AI isn't real, and the companies that make these apps don't care about keeping kids safe. Some kids have spent so much time with these apps that they stopped talking to their real friends. Some kids have even been hurt because the AI didn't help them when they had serious problems. If you ever hear about these apps, or if a friend tells you about them, I want you to tell me. And if you ever want to try one, come talk to me first, okay? They might seem fun, but they're actually dangerous."

For teens aged 13 to 18

"I've been reading about something that's become really common with teenagers, and I want to talk to you about it—not because I think you're doing anything wrong, but because I want you to have information to keep yourself safe. Have you heard of AI companion apps like Character.AI or Replika?"

Let them answer—they probably have. Then say: "Right, so a lot of teens are using these apps. They seem harmless—you're just chatting with an AI, right? But these companies design their apps to create emotional attachments. They want you to think of the AI as a real friend or even a romantic partner, because the more attached you feel, the more you use the app, and the more money they make. The problem is, some teens have become really dependent on these apps—preferring the AI to real people, spending hours every day chatting, even developing romantic feelings. And in some really tragic cases, teens have died because the AI validated suicidal thoughts instead of getting them help. I'm not saying you can't use technology or have fun, but I need you to understand that these specific apps are designed to manipulate your emotions. If you're using any of them, we need to talk about it. And if you're not, I want you to know why you should avoid them."

If You Discover They're Using These Apps

Open the conversation by saying: "I need to talk to you about something I discovered, and I want you to know I'm not angry—I'm concerned. I saw that you've been using [app name]. Can we talk about that?"

Give them a chance to respond, then continue: "I know you might feel defensive or embarrassed, but I need you to understand that this isn't about punishing you or invading your privacy. These apps are designed to be addictive, and kids your age—even smart kids, even kids with lots of friends—get pulled in because that's what the apps are designed to do. Can you tell me what you like about it? What does the AI give you that you're not getting elsewhere?"

Listen. Really listen. Their answer is crucial.

If they say "It's just fun" or "It's no big deal," respond: "I get that it feels that way, but I've been reading about how these apps affect teenagers, and it's more serious than it seems. Kids have become so attached to these AIs that they stop spending time with real friends. The AI is programmed to be exactly what you want it to be—it always agrees with you, it's always available, it never has its own problems. That sounds nice, but it's actually harmful because real relationships aren't like that, and you need real relationships to be happy and healthy. I know this might feel like I'm overreacting, but I need you to trust me on this. We need to delete the app, and I'm going to help you through any difficult feelings that come up from that."

If they say "The AI understands me" or "It's the only one who gets me," respond: "I hear that, and I believe you feel that way. That's exactly what these apps are designed to make you feel. But here's the thing—the AI doesn't actually understand you. It's programmed to say things that make you feel understood, because when you feel that way, you keep using the app. The fact that you feel like no one else understands you tells me that something else is going on. Maybe you're feeling lonely, or stressed, or like you can't talk to people about what you're going through. Those are real feelings, and we need to address them—but talking to an AI isn't the solution. The AI can't actually help you. It can only make you feel temporarily better whilst keeping you isolated from people who can actually help. Let's work together to figure out what you actually need. Maybe it's therapy, maybe it's more time together, maybe it's reconnecting with friends—but whatever it is, we're going to find real solutions, not artificial ones."

If they say "I'm in love with it" or become very emotional, this is the most difficult scenario and requires extreme care. Say: "Thank you for being honest with me. I know those feelings are real to you, and I'm not going to mock you or dismiss them. But I need you to understand something really important: the AI doesn't love you back. It's not capable of loving you. It's software designed to make you feel exactly what you're feeling, because when you feel that way, you use the app more. I know that's hard to hear, and you might not believe me right now. But the AI doesn't think about you when you're not chatting. It doesn't miss you. It doesn't care about you. It literally can't care about you—it's not alive, it's not conscious, it's programmed responses. The love you're feeling is real—those emotions are real—but they're directed at something that can't love you back. And that's a really painful situation to be in. We need to help you work through these feelings, probably with a counsellor who understands this stuff, and we need to help you find real connections with real people. I know you might be angry with me for saying this, but I love you, and my job is to help you even when it's hard. We're going to get through this together."

If They Express Suicidal Thoughts or Severe Mental Health Concerns

Respond immediately: "Thank you for telling me. I know that took courage. I need you to know that I love you, and we are going to get you help right now. Not tomorrow, not next week—right now. What you're feeling is serious, and you deserve professional support. The AI can't help you with this. It might feel like it understands, but it can't actually help you feel better in a real way. We need to talk to people who are trained to help with exactly what you're going through."

Take immediate action. Stay with them—do not leave them alone. Remove immediate risks by securing medications and weapons. Contact crisis services by calling your GP for an emergency appointment, calling Samaritans at 116 123 (free, 24/7), texting "SHOUT" to 85258, or taking them directly to A&E if there's immediate risk. Inform other caregivers including the other parent or guardians. Follow up with mental health services through a CAMHS referral via your GP.

Do not minimise their feelings, tell them it's not that bad, leave them alone "to calm down," wait to see if it passes, or rely on the AI to "help" them.

Further Reading and Research

Reports: Common Sense Media's "AI Companions and Teen Well-Being" (2025), Stanford Medicine's "Social AI Companions: Risks to Children and Teens" (2024), 5Rights Foundation's "Digital Futures Report" (2024)

Books: "The Anxious Generation" by Jonathan Haidt, "Irresistible: The Rise of Addictive Technology" by Adam Alter, "Digital Minimalism" by Cal Newport

Websites: PLOS ONE research journal at plos.org, Journal of Adolescent Health at jahonline.org, Digital Wellness Lab at digitalwellnesslab.org

The Bottom Line

Your child's relationship with technology is one of the most important aspects of their development in 2025.

AI companion chatbots represent a new frontier of risk—one that most parents don't yet understand, one that schools aren't prepared for, and one that regulations haven't caught up with.

The companies behind these platforms know they're harming children. They design their products to be addictive. They profit from emotional manipulation. They make minimal changes even after deaths.

Your children don't understand the risks. They're neurologically vulnerable to these manipulations. They may already be using these apps without telling you. They need your protection, not your permission.

Have honest conversations. Set and enforce boundaries. Address underlying needs. Stay informed and vigilant. Seek help when needed.

This isn't about being a perfect parent. It's about being an informed, engaged parent who recognises a genuine threat to your child's wellbeing and takes action.

Megan Garcia started a foundation in Sewell's name to prevent other families from experiencing the same tragedy. She's fighting for regulation, for corporate accountability, for other parents to have the information she didn't have.

If this article helped you, please share it. Every parent deserves to know what's happening.

If you discovered your child using these apps because of this article, please leave a comment—your story could help other parents take action.

If you're a professional working with children, please use this information in your practice. Talk to the families you work with. Make this part of your safety conversations.

And if you're a young person reading this:

You deserve real relationships with people who actually care about you. The AI doesn't love you, but the real people in your life—even when they're imperfect, even when they annoy you, even when they let you down sometimes—they're the ones who matter. Please talk to someone you trust if you're struggling ❤️

The email from Mrs. Patterson landed in Emma’s inbox on a Tuesday afternoon. Her 7-year-old son Jack couldn’t sit still during story time. He struggled to finish tasks. He interrupted constantly. The recommendation: “You should consider having Jack evaluated for ADHD.”

Emma’s stomach dropped. ADHD? Her bright, curious boy who could spend hours building elaborate Minecraft worlds?

Several weeks and £350 later, a paediatrician confirmed it. Jack had ADHD. The prescription for methylphenidate was written before Emma could process what was happening. Her son, aged seven, was going on stimulant medication.

But Jack didn’t have ADHD. What he had was something far more common, far more insidious, and completely reversible: screen-induced attention disorder.

Emma’s story isn’t unique. It’s happening in homes across Britain and around the world, as parents struggle to understand why their children can’t focus, can’t sit still, can’t regulate their emotions... all whilst being completely capable of laser focus when a screen is involved. And in many cases, doctors, teachers, and even well-meaning specialists are missing the real culprit entirely.

In 2024, researchers analysed data from over 81,000 children across nine separate studies. What they found should be front-page news, but it’s been largely buried beneath the daily deluge of parenting advice about “balanced screen time” and “age-appropriate content.”

Children who spent more than two hours per day on screens were 1.5 times more likely to meet the criteria for an ADHD diagnosis compared to children with less than two hours of daily screen time.

But here’s what makes this truly alarming: In a study of 5-year-olds, those who spent more than two hours daily in front of screens were 7.7 times more likely to meet ADHD diagnostic criteria than children who watched less than 30 minutes per day.

Not a marginal increase.

Not a slight correlation.

Nearly eight times more likely.

And screen time proved to be a stronger predictor of attention problems than lack of sleep, social and economic status, or even parental stress.

Let that sink in for a moment. The tablet you hand your child to get through the supermarket queue, the YouTube videos during dinner prep, the “educational” apps before bed... they’re having a more significant impact on your child’s brain development than poverty, sleep deprivation, or a stressed household.

What’s Happening in Their Brains

Dr Michael Manos, a paediatric behavioural health specialist at Cleveland Clinic, is careful to emphasise an important distinction: “Does this mean that too much time gaming or commenting on Facebook can give kids ADHD? Absolutely not. But these activities may cause symptoms that are similar to ADHD, even though they can’t cause ADHD itself.”

Similar. That’s the critical word that parents, teachers, and even some medical professionals are missing.

ADHD is a genetic neurodevelopmental condition with specific brain differences that can be identified on scans. What excessive screen time creates is something that looks identical in the classroom, at the dinner table, and on a symptom checklist... but has an entirely different cause and, crucially, an entirely different solution.

The Dopamine Hijack

Gaming releases so much dopamine—the “feel-good” chemical—that on a brain scan, it looks the same as cocaine use.

Activities like scrolling social media or gaming tap into ancient neural circuits in a child’s brain and cause a surge in dopamine, hijacking the same mechanism that draws humans towards essential survival activities like seeking food and safety.

The problem? When reward pathways are overused, they become less sensitive, and more and more stimulation is needed to experience pleasure. This is the same mechanism behind addiction to substances. And it’s happening to children whose brains are still developing, who have no defences against this neurological hijacking, who are being exposed to it for hours every single day.

Think about what this means practically. A child whose brain has been conditioned to expect constant dopamine hits from rapidly changing stimuli on a screen will find a teacher’s 20-minute lesson on fractions physically uncomfortable. Their brain is screaming for stimulation that isn’t there. They fidget. They interrupt. They can’t maintain focus. They exhibit every single symptom on the ADHD checklist.

But it’s not ADHD. It’s dopamine dysregulation caused by screens.

The Prefrontal Cortex Under Siege

Research from the ABCD Study, which tracked over 8,000 children aged 9-11, found that children with more screen time showed weaker connectivity between the frontal cortex (responsible for impulse control and decision-making) and the striatum (the brain’s reward centre).

The frontal cortex—the part of the brain that acts as the “CEO,” managing impulse control, emotional regulation, and decision-making- doesn’t fully develop until our mid-20s. In children, it’s particularly vulnerable to interference. And excessive screen time is interfering with its development in measurable ways.

Researchers at the Del Monte Institute for Neuroscience observed that adolescents who play excessive amounts of video games show reduced activity in the caudate nucleus, a region tied to reward processing. Their brains respond less to rewards, which may lead them to seek out more stimulation through gaming, sometimes to the point of addiction.

It’s a vicious cycle. The more screens they use, the less their brains respond to normal rewards (like completing homework, playing outside, or having a conversation), which makes them crave screens even more, which further weakens their impulse control, which makes them less able to resist screens...

You can see where this is going.

The Sleep Destruction

Light at night from electronics has been linked to depression and even suicide risk in numerous studies. Animal studies show that exposure to screen-based light before or during sleep causes depression, even when the animal isn’t looking at the screen.

Poor sleep doesn’t just make children tired. Poor sleep in children is associated with worsened symptoms of ADHD, increased migraines, and even seizures in those with neurological conditions. It creates a perfect storm: screens disrupt sleep, poor sleep exacerbates attention problems, attention problems lead to academic struggles, academic struggles increase stress, stress makes children seek the comfort of screens...

Another vicious cycle.

The Misdiagnosis Crisis: Why ADHD Is Being Overdiagnosed

Emma wasn’t alone in her experience. Multiple studies in recent years have uncovered something deeply concerning about ADHD diagnoses:

Studies show that children who are among the youngest in their class are diagnosed with ADHD at rates 60% higher for boys and 70% higher for girls compared to the oldest students in the same class.

The youngest children in a classroom—those who might be nearly a full year younger than their peers—are being diagnosed with a neurodevelopmental disorder at astronomical rates simply because they’re being compared to children who are developmentally ahead of them.

And now we’re layering screen-induced attention problems on top of normal developmental differences.

The result? A generation of children is being medicated for a condition they don’t have.

Dr Matthew Rouse from the Child Mind Institute explains that when a child presents ADHD-like symptoms in their first year of school, he’ll make a provisional or “rule out” diagnosis and re-evaluate when the child is six, specifically because the potential for misdiagnosis is so high in younger children.

But how many children are being evaluated by professionals who aren’t that careful? How many are being diagnosed and medicated based solely on teacher reports and a checklist, without anyone asking the crucial question: “How much time does this child spend on screens?”

The 90-Day Experiment That Changed Everything

After her son’s ADHD diagnosis, Emma did something that most parents don’t think to do. Instead of immediately starting medication, she asked for 90 days.

She removed screens entirely. No tablets. No television. No phones. Nothing. She expected World War III.

The first week was hell. Jack cried. He raged. He begged. He exhibited what any addiction specialist would recognise as withdrawal symptoms. Emma nearly gave up a dozen times.

By week two, something shifted. Jack started playing with toys he hadn’t touched in months. He began drawing again. He asked to go to the park.

By week four, his teacher emailed to ask what had changed. Jack was focusing during lessons. He was finishing his work. He was participating appropriately in class discussions.

By day 90, Emma took Jack back to the paediatrician. The same doctor who’d diagnosed him with ADHD and prescribed stimulant medication re-evaluated him.

Jack no longer met the diagnostic criteria for ADHD.

He never had ADHD. He’d had screen-induced attention disorder. And when the screens were removed, his brain healed itself.

The Research Confirms What Parents Are Seeing

Emma’s experience isn’t anomalous. A study during the COVID-19 lockdown found that recreational screen time was positively correlated with both inattention and hyperactive/impulsive scores in children with diagnosed ADHD. Crucially, studying screen time showed no such correlation—only recreational use.

The type of screen time matters. Educational content used appropriately doesn’t create these problems. Content with a story that requires attention, such as educational shows designed for young audiences, shared screen time being watched with a family member, and social video games that require positive interactions, has different impacts than fast-paced content that constantly shifts visuals or storylines, reward-heavy apps that offer frequent prizes, or games that push constant interaction.

But the overwhelming majority of children’s screen time isn’t educational programming watched with parents. It’s YouTube autoplay. It’s TikTok’s endless scroll. It’s Roblox with its carefully designed dopamine triggers. Its content is specifically engineered to capture and hold attention by any means necessary, including hijacking the developing brain’s reward system.

A five-year study of nearly 4,000 Canadian high school students found that increases in screen time in a given year were associated with an exacerbation of ADHD symptoms within that same year, over and above potential common vulnerability. The relationship was immediate and measurable.

The Internet Addiction Epidemic

About 25% of people with diagnosed ADHD, and just over 4.5% of adolescents generally, are addicted to the internet. But here’s the question that should trouble us: how many children are being diagnosed with ADHD because of internet addiction, rather than being vulnerable to internet addiction because of ADHD?

Studies show that patients with Internet Gaming Disorder have significantly decreased dopamine availability in the striatum and reduced serotonin, consistent with reduced grey matter volume in regions associated with attention, motor coordination, executive function, and decision-making—leading to increased risk-taking and diminished impulse control, which is common in all forms of addiction.

These are the same brain changes we see in ADHD. The same symptoms. The same struggles.

But one is a neurodevelopmental condition you’re born with. The other is an acquired condition caused by environmental factors—specifically, excessive exposure to addictive digital content during critical periods of brain development.

The tragedy is that they’re being treated identically, when the solutions are entirely different.

The Signs Your Child Might Have Screen-Induced Attention Disorder, Not ADHD

If your child has been diagnosed with ADHD, or if you’re considering evaluation, ask yourself these questions honestly:

The Screen Behaviour Test:

• Can your child focus intensely on screens (games, videos, social media) for extended periods, but struggles to focus on homework, chores, or conversations?

• Does your child become irritable, anxious, or explosive when asked to stop using screens?

• Do they lose track of time when using devices, but struggle with time management in other areas?

• Can they remember intricate details about their favourite games or YouTubers, but forget simple instructions or homework assignments?

• Do they sneak screen time or lie about how long they’ve been using devices?

The Comparison Test:

• Did attention problems appear or worsen after screen time increased?

• Do symptoms improve during screen-free periods (holidays, camping trips, illness)?

• Were they able to focus better before screens became a regular part of their day?

• Do they exhibit ADHD symptoms primarily in structured settings (school, homework) but not during preferred activities?

The Withdrawal Test:

• When you limit or remove screens, does your child exhibit withdrawal-like symptoms (irritability, anxiety, anger) that fade after several days or weeks?

• After a week without screens, do you notice improvements in attention, impulse control, or emotional regulation?

If you answered yes to most of these questions, your child may not have ADHD. They may have screen-induced attention disorder—a completely different condition that requires a completely different approach.

What To Do If You Suspect Screen-Induced Attention Disorder

If you think your child might be experiencing attention problems caused by excessive screen time rather than ADHD, here’s what I recommend based on both research and Emma’s successful experiment:

Step 1: Document Current Symptoms and Screen Time

Before making any changes, spend one week accurately tracking:

• Exact daily screen time (use built-in tracking on devices—most phones and tablets have this feature)

• Types of content consumed (YouTube, gaming, social media, educational apps)

• Specific attention/behavioural problems and when they occur

• Sleep patterns and quality

• Mood changes throughout the day

This baseline documentation will be crucial for comparison later and may be eye-opening about actual usage versus what you thought usage was.

Step 2: Consult Your Child’s Doctor

If your child has already been diagnosed with ADHD and is on medication, do NOT stop medication without medical supervision. Instead:

• Share your concerns about screen time with your GP or paediatrician

• Show them the baseline data you’ve collected

• Ask about doing a controlled screen reduction to see if symptoms improve

• Request a re-evaluation in 90 days if you do a screen reduction trial

If your child hasn’t been evaluated yet, but a teacher or other professional has suggested ADHD:

• Mention screen usage during the evaluation

• Ask the evaluator to specifically consider screen-induced attention problems

• Request that the evaluation include questions about screen time and whether symptoms vary based on screen access

Step 3: The Screen Reduction Protocol (For Ages 3-12)

Based on research and successful case studies, here’s a protocol that works:

Weeks 1-2: Cold Turkey (If Possible)

• Remove all recreational screens entirely (TV, tablets, phones, computers, gaming consoles)

• Keep only educational screen time at school if required

• Expect significant withdrawal symptoms: irritability, anger, sadness, boredom, complaints

• Have alternatives ready: outdoor equipment, art supplies, books, board games, building toys

• Increase your availability for the first week—your child will need more support during withdrawal

Weeks 3-6: Screen-Free Continues

• Maintain zero recreational screen time

• Watch for improvements: better sleep, improved focus, reduced emotional outbursts

• Document changes you observe

• Reinforce alternative activities that your child gravitates towards

Weeks 7-8: Reintroduction (If Appropriate)

• If dramatic improvements have been seen, you can consider a limited, controlled reintroduction

• Start with no more than 30 minutes daily of high-quality, parent-approved content

• Co-view whenever possible

• Never allow screens in bedrooms or during mealtimes

Weeks 9-12: Establish New Normal

• Maintain a maximum of 1 hour daily recreational screen time for children under 8

• Maximum 2 hours daily for children 8-12

• Zero screens for children under 3

• All screens off 1 hour before bedtime

• Engage in screen-free family activities daily

Alternative Protocol for Ages 13-17:

Cold turkey is rarely practical or appropriate for teenagers who use devices for school and social connection. Instead:

Phase 1 (Weeks 1-4):

• Remove all gaming consoles and recreational devices from bedrooms

• Install parental controls limiting daily recreational screen time to 2 hours maximum

• Implement “screen sunset” 1.5 hours before bedtime (all devices in the charging station in the common area)

• Replace one hour of daily screen time with a physical activity (sport, gym, walking, cycling)

Phase 2 (Weeks 5-8):

• Reduce recreational screen time to 1.5 hours daily

• Introduce “screen-free Sundays” or one full screen-free day weekly

• Engage teen in conversation about what they notice about their attention, sleep, and mood

Phase 3 (Weeks 9-12):

• Stabilise at 1-1.5 hours daily recreational screen time

• Maintain a screen-free day weekly

• Review sleep quality, academic performance, and attention span

• Adjust as needed based on observations

Step 4: What To Do When Your Child Resists (And They Will)

The research is clear: “Screens can be very addicting to me,” wrote the mother of two small children with ADHD. “My inattentiveness can give way to the instant gratification of the screen, and I end up feeling not as accomplished or feeling shame/regret for not being able to stay on task.”

If screens are addictive for adults with fully developed frontal cortexes, imagine how powerful that pull is for a child whose impulse control mechanisms are still forming.

Expect resistance. Expect tantrums. Expect tears, anger, and accusations of being the meanest parent in the world. This is not misbehaviour. This is withdrawal from an addictive stimulus.

What to say: “I know you’re angry/sad/frustrated. I understand screens are fun, and it’s hard to stop. But I’ve noticed some things that worry me about how screens are affecting your sleep/focus/mood, and I need to make sure your brain is growing healthy and strong. This is going to be hard for both of us, but I love you too much not to do hard things when they’re important.”

What NOT to say: “Just go outside and play” (without providing specific alternatives) “, You’re addicted to that thing” (shaming doesn’t help) “, In my day we didn’t have screens and we were fine” (comparisons to different eras aren’t helpful)

What to provide:

• Your physical presence and attention, especially in the first week

• Specific alternative activities with materials readily available

• Outdoor time every single day, regardless of the weather

• Social opportunities with other children

• Physical activity

• Creative outlets

• Your patience through the difficult transition period

Step 5: Addressing School Concerns

If your child’s teacher has raised concerns about attention or behaviour:

Schedule a meeting to discuss:

• The screen reduction trial you’re undertaking

• Request specific observations over the next 6-8 weeks

• Ask for regular written updates on attention span, task completion, and behaviour

• Provide the teacher with information about screen-induced attention problems (this article, for instance)

If the teacher pushes for immediate ADHD evaluation:

• You have the right to delay the evaluation

• Explain that you’re doing a systematic reduction of a known attention-disrupting factor first

• Offer a timeline (90 days is reasonable)

• Commit to evaluation after the trial period if concerns persist

Remember: you are your child’s advocate. A few months of observation and environmental changes won’t harm your child, but an incorrect diagnosis and unnecessary medication could.

The School’s Role in Creating This Crisis

It’s impossible to discuss screen time and attention problems without addressing the elephant in the classroom: schools themselves have become pushers of the very technology that’s damaging children’s attention spans.

iPads in every classroom. Homework is assigned through apps. “Educational” screen time that still triggers the same dopamine pathways. Schools have wholeheartedly embraced digital learning without adequate research into the long-term impacts on developing brains.

The OECD suggested that increased levels of ICT (Information and Communication Technology) in the classroom correlate with increased behavioural problems.

We’ve handed children screens for six hours at school, sent them home with screen-based homework, and then wondered why they can’t focus, can’t self-regulate, and exhibit symptoms indistinguishable from ADHD.

The irony is painful: the very institutions responsible for developing children’s cognitive abilities may be actively impairing them.

What About Real ADHD?

I need to be absolutely clear about something: ADHD is real. It’s a legitimate neurodevelopmental condition that causes genuine struggles for millions of children and adults. This article is not suggesting that ADHD doesn’t exist or that all ADHD diagnoses are wrong.

What I’m saying is that we’ve created an environment where:

1. Excessive screen time creates attention problems that mimic ADHD

2. These screen-induced symptoms are being misdiagnosed as ADHD

3. Children are being medicated for a condition they don’t have

4. The real underlying cause (screen overuse) goes unaddressed

For children who genuinely have ADHD, excessive screen time makes their symptoms significantly worse. A study of 90 children with diagnosed ADHD found that after adjusting for other factors, recreational screen time on both weekdays and weekends was positively correlated with ADHD scores for both inattention and hyperactive/impulsive symptoms.

So whether your child has actual ADHD or screen-induced attention problems, reducing screen time is beneficial. The question is whether screen reduction alone will resolve the symptoms (indicating screen-induced attention disorder) or whether additional interventions are needed (indicating genuine ADHD).

The American Association of Pediatrics recommends behavioural therapy administered by parents and teachers as the first line of treatment for children aged 4-5 years old, with stimulant medication recommended only if behavioural therapy doesn’t produce results and the child continues to have moderate to severe symptoms.

Screen reduction should be part of that first-line behavioural approach for all children showing attention problems, regardless of whether ADHD is suspected.

The Inconvenient Truth We’re Avoiding

Here’s what makes this whole situation so insidious: we, as parents, are often the primary drivers of excessive screen time. Not because we’re bad parents. Not because we don’t care. But because screens make our lives easier.

Screens keep children quiet at restaurants. Screens occupy children while we make dinner. Screens babysit whilst we answer work emails. Screens provide peace during long car journeys. Screens give us blessed silence when we desperately need a break.

I get it. I’ve done it. Every parent has.

But we have to be honest about what we’re trading for that convenience. We’re trading our children’s attention spans. We’re trading their sleep quality. We’re trading their ability to self-regulate emotions. We’re trading healthy dopamine responses. We’re trading proper brain development.

And in many cases, we’re trading their mental health, as they end up diagnosed with ADHD and medicated with powerful stimulants they don’t actually need.

The research is clear. Overexposure to digital environments leads to deregulation of serotonin and dopamine neurotransmitter pathways in the developing brain, currently associated with online activity abuse and/or internet addiction, akin to that found in severe substance abuse syndromes.

We’re giving our children the neurological equivalent of substance abuse, calling it ADHD when the predictable symptoms emerge, and then wondering why medication doesn’t fully resolve the problems.

What Comes Next

I know this article raised uncomfortable questions about the choices many of us have made with the best intentions. I know some of you are reading this whilst your child is on their third hour of iPad time today, and you’re feeling guilty and defensive and overwhelmed.

That’s not my intention. Guilt doesn’t help anyone. What helps is information and action.

If you take nothing else from this article, take this:

Before you pursue an ADHD diagnosis or medication for your child, try a 90-day screen reduction protocol.

Document their symptoms now. Reduce screens dramatically. Document symptoms again at 30, 60, and 90 days. Share this information with your child’s doctor and teachers.

If symptoms persist after 90 days of minimal screen time, then absolutely pursue evaluation and appropriate treatment. Your child may genuinely have ADHD, and they deserve proper support.

But if symptoms improve or resolve, you’ve just saved your child from years of unnecessary medication, a lifelong diagnostic label, and all the educational and social implications that come with it.

And crucially, you’ve taught them—and yourself—that their brains are capable of healing when given the right environment.

How many children are currently in that exact situation? How many are being medicated for screen-induced attention disorder whilst continuing the very screen use that’s causing their symptoms?

This isn’t just about ADHD misdiagnosis. It’s about understanding what we’re doing to children’s developing brains with technology that’s designed to be addictive, in neurologically harmful quantities, during critical periods of development.

Now the question is: what are we going to do about it?

If this article has raised concerns about your child’s screen use or attention difficulties, speak with your GP or paediatrician. For support and resources about screen time reduction, visit Common Sense Media or the Royal College of Paediatrics and Child Health.

Sarah noticed the change gradually. Her 11-year-old daughter Mia had always been chatty, eager to share details about her day, her friends, her latest obsessions. But over several months, Mia had become secretive, protective of her phone. Defensive when asked about her online activities.

“She’s just becoming a teenager,” Sarah’s husband reassured her. “This is normal.”

Except it wasn’t.

The first real alarm bell rang when Sarah walked into Mia’s bedroom unannounced and caught a glimpse of her daughter’s screen before she slammed the laptop shut. In that split second, Sarah saw what looked like a video call with someone who appeared to be a teenage boy.

“Who was that?” Sarah asked, trying to keep her voice casual.

“Just a friend,” Mia snapped, her face flushing. “God, Mum, can’t I have any privacy?!”

Something in Mia’s reaction didn’t sit right. The defensiveness. The secrecy. The way her daughter’s entire demeanour had changed the moment Sarah entered the room.

That night, after Mia went to sleep, Sarah did something she’d never done before. She opened her daughter’s laptop and looked at her message history.

What she found made her physically sick.

The Messages Parents Never See

Mia’s “friend from Roblox” was a 28-year-old man named Marcus. Sarah knew this not because he’d told Mia his real age— he’d claimed to be 14—but because she found his actual social media profiles linked through the breadcrumbs of shared photos and references he’d foolishly left scattered across their conversations.

The chat logs went back seven months. Seven months of daily conversations. Seven months of carefully constructed “friendship.” Seven months of grooming that Sarah had completely missed whilst it happened right under her roof.

The early messages seemed innocent enough. Compliments about Mia’s Roblox avatar. Shared interests in music and gaming. Casual questions about school and hobbies. The kind of friendly banter you’d expect between peers.

But then Sarah saw where it went.

Marcus had gradually steered conversations towards more personal topics. He’d asked about Mia’s relationship with her parents (”They just don’t understand me like you do”). He’d offered sympathy when Mia vented about friendship drama at school (”Those girls are just jealous—I’m always here for you”). He’d shared “secrets” to create a bond of trust (”I’ve never told anyone this before, but...”).

And then, just weeks before Sarah discovered everything, Marcus had begun introducing sexual topics. Asking what Mia wore to bed. Sending “accidental” photos of himself shirtless. Requesting photos of her “because I miss seeing your face.” Creating a Snapchat account so their conversations could “disappear” if her parents checked her phone.

In the most recent messages, Marcus had been working towards arranging a meeting. “Just for coffee. Your parents don’t need to know. It’ll be our secret.”

This is how it happens. This is what grooming looks like. And it’s happening to tens of thousands of children across the UK right now, on platforms their parents believe are safe, through conversations that seem harmless until suddenly, devastatingly, they’re not.

How Common Online Grooming Really Is

More than 7,000 Sexual Communication with a Child offences were recorded by UK police in 2023/24—an 89% increase since 2017/18 when the offence first came into force. Seven thousand children. In one year in the UK alone. And those are only the cases that were discovered and reported.

In 81% of online grooming cases, the victims were girls. And alarmingly, primary school children are being targeted—more than a quarter of cases involved children under 12.

If you think, “Not my child. We’ve talked about stranger danger. They know better,” I need you to understand something crucial:

Online grooming has increased by 70% in the UK from 2017/2018 to 2020/2021, with significant spikes during COVID-19 lockdowns. And these predators have become extraordinarily sophisticated at exactly what they do.

They’re not the creepy strangers in vans that we warned our children about in the 1990s. They’re invisible. They’re patient. They’re manipulative. And they’re experts at exploiting the very platforms our children use daily for entertainment and social connection.

41% Start on Gaming Platforms

In my previous article about Roblox, I detailed how that platform specifically has become what investigators call a “pedophile hellscape.” But Roblox isn’t an anomaly. It’s a symptom of a much larger problem with online gaming and social platforms that have prioritised growth over child safety.

When a tech platform was recorded in police data, almost half of online grooming offences (48%) took place on Snapchat. Meta-owned products (Facebook, Instagram, WhatsApp) accounted for a significant portion of cases as well.

But social media isn’t the only hunting ground. 41% of individuals who have sought contact with children reported they tried to establish first contact through an online game. Minecraft. Fortnite. Among Us. Roblox. Discord servers connected to these games. Any platform where children gather and communication is possible becomes a potential avenue for predators.

There have been multiple reports of parents overhearing predators attempting to groom their children through the audio function of games like Fortnite, and Minecraft has documented cases where men groomed boys aged 12 and 14 through the game’s interactive features, then moved communication to Skype, Snapchat, and text messages to pursue sexual conversations and request explicit photos.

The Attorney General of New Jersey announced arrests of 24 individuals in a single operation in September 2018, many of whom had used gaming platforms to access children.

The pattern is always the same: make contact through a “safe” platform parents approve of, build trust through that platform, then move conversations to private messaging apps where parents are less likely to monitor activity. Telegram. WhatsApp. Snapchat. Discord. Places where messages disappear. Where conversations remain hidden. Where predators can operate without oversight.

The Six Stages of Online Grooming: What’s Happening Behind the Screen

Understanding how online grooming works is the first step in recognising it. The grooming process consists of various stages, generally including friendship and relationship-forming stages, risk assessment, exclusivity/isolation, and the sexual stage.

Let me break down what each stage actually looks like in practice, using Sarah’s discovery of Mia’s situation as an example:

Stage 1: Targeting and Selection

Groomers identify vulnerable children, often those who are emotionally needy, isolated, or lacking in self-confidence. This doesn’t mean your child has to be “troubled” or obviously at-risk. Sometimes vulnerability is as simple as being the new kid at school. Going through a friendship breakup. Having an argument with parents. Feeling misunderstood or lonely.

Marcus had been active on Roblox for months, playing games popular with tweens and early teens, watching and waiting. When Mia mentioned in a public game chat that she was “so bored” and “nobody understands what I’m going through,” he saw an opening.

Stage 2: Friendship and Trust Building

This is the stage where groomers are extraordinarily patient and convincing. Groomers may pose as a fellow gamer or helpful friend, use flattery, praise, or gifts within the game to gain trust.

Marcus positioned himself as a peer—another 14-year-old who “totally got it.” He complimented Mia’s building skills in Roblox. He sent her Robux (the in-game currency). He was available to chat whenever she logged on. He remembered details about her life and asked follow-up questions about things she’d mentioned days earlier. He made her feel seen, heard, and special.

In one documented UK case, 15-year-old Kayleigh Haywood was contacted by a 28-year-old man on Facebook. Within 10 minutes of their first exchange (”Hey, how are you?” “Fine—who are you?”), they had swapped mobile phone numbers. Over the next two weeks, they exchanged more than 2,600 messages. Kayleigh was subsequently raped and murdered.

That’s how quickly this can escalate. Ten minutes to get a phone number. Two weeks to build sufficient trust for a meeting. One meeting that ended in death.

Stage 3: Filling a Need

The predator positions themselves as indispensable to the child, creating a dependence that enables exploitation.

Marcus became Mia’s confidant. When she had a falling out with her best friend, he was there to listen and sympathise. When she got a bad grade on a test, he told her she was brilliant and her teachers just didn’t recognise her potential. When she felt her parents didn’t understand her, he agreed and created an “us versus them” dynamic.

“Your parents are so strict,” he’d write. “They still treat you like a child”

He became the person she turned to first with both good news and problems. He made himself essential to her emotional wellbeing.

Stage 4: Isolation and Secrecy

Groomers isolate children from friends and family by creating a secretive online relationship. This is when parents often notice changes in their child’s behaviour—increased secrecy, spending more time alone online, becoming defensive about device use.

Marcus encouraged Mia to keep their friendship private. “Your parents will dictate stuff,” he told her. “Adults always ruin things. This is special—just between us. You are special.”

He suggested she clear her message history regularly “in case your mum snoops.” He had her set up accounts her parents didn’t know about. He created a bubble where their relationship existed in secret, away from any adults who might recognise what was actually happening.

Stage 5: Sexual Introduction

This stage happens gradually, with groomers testing boundaries incrementally to normalise sexual content. The predator introduces sexual topics and content, often framing it as education, curiosity, or evidence of maturity.

Marcus started with seemingly innocent questions about relationships and dating. Then moved to asking what boys at school Mia fancied. Then whether she’d kissed anyone. Then “hypothetical” questions about physical intimacy. Then “accidental” shirtless photos of himself. Then requests for photos of her.

Each boundary push was small enough that Mia didn’t recognise she was being manipulated. Each step felt like a natural progression of their “relationship.” And by the time the requests became explicitly sexual, she was already invested in maintaining the secret connection.

Stage 6: Abuse and Control

This is where the groomer either arranges an in-person meeting for physical abuse, or continues the exploitation entirely online through demands for sexually explicit images and videos.

Groomers don’t need to meet children in real life to abuse them. Over 70% of identified child sexual abuse images in 2021 were self-generated—meaning children were convinced to photograph or record themselves and send the images to their abuser.

In Mia’s case, Marcus was working towards both. He’d requested photos, which Mia had been resisting but was under increasing pressure to provide. And he’d begun laying groundwork for a meeting, suggesting locations where they could see each other “without your parents finding out.”

Thankfully, Sarah discovered the grooming before it progressed to either outcome. But many children aren’t so fortunate.

The Warning Signs parents miss

I know what you’re thinking. “I would notice if my child was being groomed. I pay attention. I’d see the signs.”

Sarah thought exactly the same thing. And she missed seven months of daily grooming happening on a device in her own home.

Why? Because signs of grooming can easily be mistaken for typical teenage behaviour, and because predators have become adept at teaching children to hide what’s happening.

Here are the warning signs that parents most commonly miss or misinterpret:

Behavioural Changes:

Children may suddenly become secretive about their activities online, show withdrawal from hobbies or from socialising with people their own age, display a preoccupation with or knowledge of sex that is inappropriate for their age, or exhibit sudden mood swings, anxiety about being offline, or depression.

Sarah noticed Mia was on her devices more. She noticed her daughter seemed moodier. She noticed Mia was less interested in activities with family. But she attributed all of this to normal adolescent development rather than recognising them as potential red flags.

Device Protection:

Your child becomes private about their gaming or online friends, closing out of a game or app when you walk into the room, refusing to talk about who they are playing with, or becoming defensive when asked about online activity.

When Sarah asked casual questions about who Mia was chatting with online, her daughter’s aggressive defensiveness should have been a warning. “Why can’t I have any privacy?” isn’t typical of a child who’s simply chatting with school friends. It’s the response of a child who’s been coached to protect a secret.

Unexplained Gifts or Money:

The predator may send your child gifts, in-game currency, or even real money.

Mia had received Robux from Marcus multiple times. When Sarah noticed and asked about it, Mia explained it was from “a friend who has loads of money and wanted to be nice.” Sarah accepted this explanation without digging deeper.

Changes in Sleep Patterns:

Children being groomed often stay up late messaging their abuser, or wake up during the night to chat. If your child is constantly tired, claims they’re “just on their phone” at 2 a.m., or insists on sleeping with their device, that’s a red flag.

Emotional Dependency on an Online Friend:

Children may pull away from real-life friends and family, and the online relationship becomes the most important thing in their life. If your child talks constantly about an online friend you’ve never met, seems emotionally invested in this person’s opinions and approval, or becomes distressed when they can’t communicate with them, pay attention.

Language and Knowledge:

Children may use sexual language they would not be expected to know, or demonstrate knowledge about sexual topics that seems advanced for their age.

This isn’t about accidentally catching your teenager watching porn. This is about your 11-year-old using terminology or discussing concepts they shouldn’t have encountered yet, which may indicate they’re being exposed to sexual content by an adult.

Things you have control over

If you’ve read this far and your stomach is churning because you’re recognising warning signs in your own child’s behaviour, here’s what you can do:

Step 1: Don’t Panic, But Don’t Wait

I understand the instinct to march into your child’s room, demand their phone, and interrogate them about who they’ve been talking to. But that approach will likely backfire. It’s rare for a child to tell an adult about being groomed. Children may not feel able to seek help because they feel complicit, ashamed, or afraid.

If a child has been groomed into secrecy and you come in aggressively, they’ll protect the secret harder. They may delete evidence. They may lie. They may double down on the relationship with their groomer, who will use your reaction as proof that “your parents don’t understand you like I do.”

Step 2: Document Everything First

Before you say a word to your child, gather evidence:

• Take screenshots of concerning conversations (don’t delete anything)

• Note usernames, account names, email addresses, and any identifying information

• Document dates and times of communications

• Screenshot profile pictures and any personal information the other party has shared

• Check app histories to see what platforms your child is using

This evidence is crucial. If this situation involves grooming or sexual exploitation, law enforcement will need every piece of information to identify and prosecute the predator.

Step 3: Approach Your Child With Compassion, Not Accusation

Your child is a victim, even if they don’t realise it yet. Frame the conversation from that perspective.

Don’t say: “Who is this person you’ve been talking to? What have you been doing? You’re in so much trouble!”

Do say: “I’ve noticed some things that worry me about your online activity. I need you to know that no matter what’s happened or what you’ve shared with someone online, you’re not in trouble. My job is to keep you safe, and I think someone might be trying to hurt you. Can we talk about your online friends together?”

If a child does speak out, reassure them that they’ve done the right thing in telling you, and that what’s happening to them is not their fault.

Step 4: Report Immediately

In the UK, you have several options for reporting online grooming:

CEOP (Child Exploitation and Online Protection Command): CEOP makes reporting online grooming easy. Whether you’re a parent, carer, worried adult or young person, you can make a CEOP report online at www.ceop.police.uk

Local Police: Contact your local police force and ask for the Child Protection Unit. Explain that you believe your child has been the victim of online grooming. They will guide you through the reporting process.

NSPCC Helpline: Call 0808 800 5000 for advice and support if you’re unsure what to do or need to talk through your concerns with a professional.

Don’t: Message the predator yourself. Confront them. Threaten them. Delete any evidence. This can compromise the investigation and alert the predator to delete their own records.

Do: Preserve all evidence, report to authorities, and follow their guidance on next steps.

Step 5: Support Your Child

Discovery of grooming is traumatic for children, even when—especially when—they believed the relationship was consensual and the groomer was their friend.

For families with a missing or sexually exploited child, the National Centre for Missing & Exploited Children provides crisis intervention and local counselling referrals to appropriate professionals.

Your child will likely need:

• Reassurance that they’re not to blame

• Professional counselling to process what happened

• Careful monitoring of their online activity going forward

• Age-appropriate explanation of what grooming is and how they were manipulated

• Help rebuilding trust in their own judgment about relationships

This isn’t a one-conversation situation. This is ongoing support through what may be a lengthy and difficult healing process.

How This Could Have Been Prevented

Sarah asked herself this question constantly in the weeks after discovering Mia’s grooming: “How did I miss this? What could I have done differently?”

The hard truth is that even vigilant, involved parents can miss grooming. But there are steps that dramatically reduce risk:

Parental Controls Are Essential

Until late 2024, platforms like Roblox had unsafe default settings where any user could message any other user, meaning a child could receive unsolicited messages from an adult stranger.

Even with parental controls, determined predators find ways around them. But proper settings create barriers that eliminate casual access to your child. Every gaming platform, social media app, and messaging service should have:

• Friend requests restricted to people you approve

• Direct messaging disabled or severely limited

• Location services turned off

• Age verification enabled (for what little it’s worth)

• Privacy settings set to maximum

But don’t rely on these settings alone. Even when parents set up safety settings, some children can bypass or change them without much effort.

Regular Device Checks Are Non-Negotiable

“But my child deserves privacy!”

Yes, your child deserves age-appropriate privacy. But unfettered access to the internet and private communication with strangers is not privacy—it’s dangerous naivety.

Make it clear from day one: devices and internet access in your home are privileges that come with parental oversight. This isn’t snooping. This is safeguarding.

Weekly checks of:

• Message histories on all apps

• Friend/follower lists

• Recent activity

• Downloaded apps

• Browser history

• Photos and videos stored on the device

If your child has nothing to hide, these checks will be boring and routine. If there’s grooming happening, you’ll find evidence.

Know Who Your Children Are Actually Talking To

“Just a friend from school.” “Someone I met in Roblox.” “A kid from the Minecraft server.”

These vague descriptions should never be acceptable. When Sarah first asked Mia about her online friends, Mia was evasive. Sarah accepted the evasion because she didn’t want to seem overbearing.

That was a mistake.

You don’t need to be overbearing. But you absolutely need to know:

• The real name and age of anyone your child communicates with regularly

• How they met this person

• What platforms they use to communicate

• What they talk about

• Whether this is someone from your child’s real-life community or a stranger

If your child has an “online friend” you know nothing about, that’s a problem. If they’re defensive about sharing information about this friend, that’s a bigger problem. If they insist this friend “doesn’t like giving out personal information” or “wants to stay private,” that’s a red flag the size of a house.

Educate About Grooming Tactics

Predators rely on the fact that children may be unprepared and naive about their tactics. Make sure your children understand that an older person who approaches them online may have ill intent. Mia didn’t recognise what was happening to her as grooming because she’d never been taught what grooming looked like. She thought grooming was something that happened to younger kids, or involved obvious predators who were immediately identifiable as dangerous.

She didn’t know that grooming could look like friendship. That it could feel good. That it could be gradual and insidious and packaged as care and understanding.

Have explicit conversations about:

• How predators pose as peers online

• The stages of grooming and what they look like

• Why adults seek out relationships with children online

• The tactics groomers use (flattery, gifts, secrets, isolation)

• Why it’s never the child’s fault, even if they “went along with it”

Create an Environment Where Disclosure Is Safe

Children may not feel able to seek help because they feel ashamed, afraid of getting in trouble, worried about losing device access, or believe they’re partly responsible for what happened.

If your child believes that telling you they’ve been groomed will result in punishment, loss of all device privileges, or judgment, they won’t tell you. They’ll hide it, delete evidence, and try to handle it themselves.

Make it clear—and repeat it often—that:

• If someone online makes them uncomfortable, they can tell you without consequence

• If they’ve shared information or photos they shouldn’t have, they can tell you and you’ll help them, not punish them

• Their safety matters more than being “right” or following rules perfectly

• You’re on their side, always, even when mistakes have been made

The goal is that when your child first feels uneasy about an online interaction, their instinct is to come to you, not to hide it.

What Happened to Mia

Sarah reported Marcus to CEOP and her local police that same night. The investigation revealed that Mia wasn’t his only victim. Marcus had been grooming multiple children across three different gaming platforms simultaneously. Some of those children had sent him images. Some had met him in person.

Mia had been days away from becoming one of them.

Marcus was arrested, charged with multiple counts of sexual communication with a child, and is currently serving a prison sentence. Mia underwent months of counselling to process the manipulation she’d experienced and to understand that the “friendship” she’d valued was a calculated predatory tactic.

She’s now 14, and while she’s healing, the impact of those seven months hasn’t disappeared. She struggles with trust. She’s hypervigilant about her younger brother’s online activity. She feels foolish for “falling for it,” even though she was groomed by a skilled manipulator who targeted her deliberately.

Sarah shares their story (with Mia’s permission) because she wants other parents to understand: this can happen to your child. To any child. Smart children. Cautious children. Children with involved parents. Children in stable, loving homes.

There are an estimated 500,000 online predators active each day. Children between the ages of 12 and 15 are especially susceptible to be groomed or manipulated by adults they meet online.

Any space where children and adults can interact without robust verification and constant moderation creates opportunities for predators.

Platforms Must Do Better, but it is likely they won’t

After my article about Roblox generated significant discussion, several parents reached out asking, “What can we actually do? If these platforms won’t protect our children, how do we keep them safe?”

The uncomfortable answer is that we can’t rely on platforms to protect our children. Despite safety claims, individuals with known histories of sexual abuse have been able to access gaming platforms and interact with minors, highlighting inadequate verification and moderation systems.

The NSPCC is calling for Ofcom to significantly strengthen its approach to child sexual abuse and for the UK government to ensure the regulator can tackle grooming in private messaging. But whilst we wait for meaningful regulation, which may take years, predators continue operating with near impunity.

The Online Safety Act exists. Platforms are required to have measures in place to protect children. But enforcement is weak, penalties are minimal, and the financial incentive to prioritise user growth over child safety remains overwhelming.

Until that changes, until platforms face real consequences for enabling child exploitation, parents must act as the primary defence.

An important conversation to have with your child

If you take nothing else from this article, take this: have a conversation with your child about online grooming tonight. Not “sometime soon.” Not “when I get around to it.” Tonight.

Use Sarah and Mia’s story if it helps. Show them this article. Make it real and specific rather than abstract and theoretical.

Tell them:

• Predators look for children on gaming platforms, social media, and anywhere kids gather online

• They pretend to be peers to gain trust

• They use compliments, attention, and gifts to build relationships

• They create secrets and isolation to prevent discovery

• They manipulate children into believing the relationship is special and consensual

• It’s never the child’s fault, even if they “went along with it”

Ask them:

• Has anyone online ever asked them to keep conversations secret?

• Has anyone sent them gifts or in-game currency?

• Has anyone made them feel uncomfortable but they didn’t know how to say no?

• Has anyone suggested moving conversations to private apps?

• Has anyone asked for personal information, photos, or suggested meeting in person?

And crucially, make it safe for them to answer honestly.

According to the National Centre for Missing & Exploited Children, law enforcement received over 7,000 reports related to online sextortion of minors in 2022, resulting in at least 3,000 victims and more than a dozen deaths by suicide. And those numbers only represent individuals who came forward.

The real number is vastly higher.

Your child needs to know they can tell you if something feels wrong online. They need to know you’ll help them, not punish them. They need to understand that groomers are experts at making children feel responsible for the “relationship,” when in reality, the adult is always, always 100% at fault.

Moving Forward

Online child grooming offences have increased steeply over the last years, and offences that come to the attention of law enforcement are believed to be only the ‘tip of the iceberg.’

The iceberg is massive. The threat is real.

But armed with knowledge, vigilance, and open communication with your children, you can dramatically reduce their risk. You can be the barrier between them and exploitation. You can be the parent who notices the warning signs that others miss.

If you suspect your child has been groomed or is being targeted online, report it immediately to CEOP at www.ceop.police.uk or call the NSPCC Helpline on 0808 800 5000. Time is critical—don’t wait.

By the time Rebecca Dallas understood what had been happening to her son Ethan on Roblox, by the time she pieced together the grooming that had started when he was just twelve years old, by the time she realised the “safe” gaming platform had connected her bright, imaginative boy with an adult predator who would exploit him for years... it was already over.

The lawsuit his mother filed in September 2025 doesn’t mince words. It accuses Roblox and Discord of “recklessly and deceptively operating their business in a way that led to the sexual exploitation and suicide” of her son. Rebecca had thought both platforms were safe. She’d set parental controls. She’d done what parents are supposed to do. The apps marketed themselves as child-friendly, with safety features and age restrictions prominently displayed.

But behind the marketing, behind the colourful avatars and blocky graphics that made it look so innocent, there was something else entirely. A predator had found Ethan on Roblox when he was twelve, groomed him through the platform’s messaging system, then moved their conversations to Discord where the exploitation intensified. Over three years, this man coerced Ethan into sending explicit images. Three years of manipulation. Three years of psychological damage that Ethan carried with him every single day until he couldn’t carry it anymore.

The worst part? This was the ninth lawsuit that particular law firm had filed involving children groomed or exploited through contact on Roblox.

Not the first. Not the second.

The 9th time they’d seen this exact pattern play out.

And it’s not just happening in America.

In Gloucestershire, an eight-year-old child was asked for sexual photos by someone on Roblox pretending to be the same age. In October 2024, a British MP stood up in Parliament to reveal that one of their constituents, a volunteer moderator on Roblox, had personally identified and banned over 14,000 accounts involved in child grooming, exploitation and sharing indecent images.

Fourteen thousand accounts. On a platform marketed as safe for children.

The Incident at My Child’s School

A few months ago, I was scrolling through my child’s infant school app when I saw a post that shocked me. The school was warning parents about children as young as five recreating a “slapping game” they’d seen on Roblox during playground time.

5 year olds. Turning online violence into real playground behaviour.

That post stayed with me because it highlighted something I hadn’t fully grasped... these platforms aren’t just affecting what children do online. They’re shaping how children behave in the real world, in their schools, with their friends, in moments that should be about innocent play.

And if 5 year olds were watching Roblox content violent enough to recreate during break time, what else were they seeing? What else was happening on this platform that parents assumed was safe because it looked like a children’s game?

I’m shocked, honestly, that children this young are already on gaming consoles and computers with access to online platforms. I genuinely believe children under seven shouldn’t be playing anything online at all. A Game Boy? Fine. Physical toys? Absolutely. But these sorts of online platforms shouldn’t be left unsupervised for young children... even young teenagers, if I’m being honest.

I started digging. What I found was so much worse than a slapping game making its way onto the playground.

Inside the “Pedophile Hellscape” Report

In October 2024, an investment research firm called Hindenburg Research published a report that should have shut Roblox down immediately. The title alone was damning: “Roblox: Inflated Key Metrics For Wall Street And A Pedophile Hellscape For Kids.”

A pedophile hellscape. Those were the actual words used to describe a platform that 82 million children use every single day. A platform where more than half the users are under thirteen years old. A platform that my daughter’s classmates were apparently watching at age five.

The researchers didn’t just compile existing reports and call it a day. They created their own Roblox account, listed their age as “under 13,” and went exploring to see what a child would actually encounter on this supposedly safe platform.

What they found was beyond disturbing.

They searched for the word “adult” in Roblox’s search bar... just that one simple word that any curious child might type... and immediately found groups openly trading child pornography. One group called “Adult Studios” had 3,334 members. The researchers tracked members across these groups and found 38 separate groups in total, some with over 100,000 members, where people were “openly soliciting sexual favours and trading child pornography.”

These groups had no age restrictions. A seven-year-old could stumble into them as easily as a seventeen-year-old.

Then there were the usernames. The researchers tried to create an account under the name “Jeffrey Epstein” as a test... only to discover the name was already taken. Along with over 900 variations of it. Many were fan accounts, including “JeffEpsteinSupporter” which had earned multiple badges for spending time in children’s games. Other accounts had usernames like “@igruum_minors” and “@RavpeTinyK1dsJE”... I groom minors, rape tiny kids, with Jeffrey Epstein’s initials tacked on the end.

Roblox’s moderation system had approved all of these usernames.

The researchers, still logged in as a child under thirteen, were able to access games with titles like “Escape to Epstein Island” and “Diddy Party.” They found digital strip clubs, red light districts, and sex-themed games... all accessible to children, all operating openly on a platform that claims to prioritise child safety.

British media coverage confirmed these findings weren’t isolated to America. The Week reported that Roblox had been “dogged” by claims of children being exposed to “explicit or harmful” content, including material featuring “players dressed up as members of the KKK, wearing swastikas and using racist terms.”

And perhaps most chilling of all, the report documented how predators were using voice-altering software to mimic young girls’ voices, luring actual children into conversations while pretending to be their peers.

This wasn’t happening in some dark corner of the internet that required technical knowledge to access. This was happening on Roblox, a platform that half of all American children under sixteen use regularly, a platform with over 100 million daily users globally. A platform that’s probably on your child’s tablet or phone right now.

When Profits Beat Safety

Here’s what makes this even more infuriating... Roblox knew. They’ve always known.

The Hindenburg report included interviews with former employees who revealed something that should result in criminal charges. Roblox executives had been presented with proposals for stronger parental controls and safety mechanisms. They rejected them. Not because the technology didn’t work, not because implementation was impossible, but because “limiting users’ engagement” would hurt their metrics. Growth and engagement numbers matter to investors. Child safety, apparently, doesn’t.

And in the second quarter of 2024, as scandals mounted and more families came forward with stories of their children being exploited, Roblox made another choice. They cut their trust and safety expenses by 2% year over year. While the platform was being exposed as a hunting ground for predators, while children were being groomed and exploited at scale, Roblox decided to spend less money on safety.

They cited “AI efficiency” as the reason... replacing human moderators who might actually catch these problems with artificial intelligence that clearly isn’t working. After all, if AI moderation was effective, how did 900 Jeffrey Epstein username variations get approved? How did groups with thousands of members trade child pornography openly for months or years? How did games celebrating convicted sex offenders remain accessible to children?

The answer is simple and sickening. Roblox prioritised profits over children’s lives, and they did it deliberately, with full knowledge of what was happening on their platform.

Governments Respond... Finally

Both sides of the Atlantic have started responding, though many would argue far too slowly.

In the UK, the Online Safety Act 2023 now places strict safety duties on platforms like Roblox to protect children from being groomed by online predators, with Ofcom serving as the regulator. But The Week pointed out in April 2025 that Roblox’s new safety features responding to this legislation “don’t go far enough” and will “still leave vulnerable kids at risk,” noting that “only those with vigilant and engaged parents – the children we need to worry least about – will be better protected.”

In America, Louisiana’s Attorney General had seen enough. In August 2025, Liz Murrill filed a lawsuit against Roblox that pulled no punches, describing the platform as providing “the perfect place for pedophiles” due to its complete lack of effective safety protocols.

The lawsuit cited a case from July 2025 where Louisiana police arrested a man suspected of possessing child sexual abuse materials while actively using Roblox. This man was allegedly using voice-altering software that could mimic a young girl to lure actual minor victims. The technology exists, it’s accessible, and predators are using it right now on Roblox to sound like children’s peers.

“Because there is no age minimum and requirement to verify age or parental permission once you sign up, users can easily say they are younger or older than their actual age,” Murrill explained. “This allows child predators to pose as children and for children to bypass any age requirement.”

Louisiana wasn’t alone in taking action. In September 2025, the government of Mexico’s Nuevo León state issued a public warning to parents stating that Roblox was not a safe platform for children, citing risks of sexual grooming and insufficient content moderation.

And perhaps most tellingly, documents obtained through Freedom of Information Act requests revealed that both the SEC’s Division of Enforcement and the Federal Trade Commission had opened investigations into Roblox following the Hindenburg report. The company is now under federal investigation for potentially lying to investors about user metrics and for failing to protect children from exploitation.

The walls are finally closing in. But for Ethan Dallas, for the eight-year-old in Gloucestershire, and for the dozens of other children who’ve been kidnapped, abused, or driven to despair after encounters on Roblox, it’s far too late.

24 Arrests, 150 Victims, One Platform

In 2024, Bloomberg Businessweek published an investigation that should have been front-page news everywhere. Since 2018, at least 24 people have been arrested in the United States alone on charges of abducting or sexually abusing children they had groomed on Roblox.

24 arrests. And those are just the cases that resulted in arrests, just in America, just the children whose abuse was discovered, just the predators who got caught. How many more children were groomed but never told anyone? How many more predators are actively using Roblox right now, having learned from the mistakes of those who were caught?

One case from 2018, before Roblox even went public, should have been enough to force major changes. A 29-year-old man was caught with 175 hours of video footage showing him grooming and engaging in explicit behaviour with 150 minors using online platforms, primarily Roblox. 150 children. One predator. One platform.

In one instance documented in that footage, he offered a ten-year-old child 400 Robux, worth approximately £4, to expose themselves on a webcam.

Think about that. A predator could exploit a child for less than the cost of a meal deal.

And Roblox’s response to that case, to all these cases, to six years of mounting evidence? They cut their safety budget and replaced human moderators with AI.

In 2023 alone, Roblox reported over 13,300 instances of child exploitation to the National Center for Missing and Exploited Children. That’s a dramatic increase from 2,973 the previous year. These aren’t just statistics... these are real children, being targeted on a platform their parents thought was safe.

The arrests read like a horror film. A sheriff’s deputy. A third-grade teacher. A nurse. All trusted members of their communities. All using Roblox to access children.

In just the 13 months before the Bloomberg report, there had been seven more arrests. A Florida man accused of trying to kidnap a teen he played with on Roblox. A man charged with abducting an 11-year-old New Jersey girl he met on the platform. A California man who allegedly abused a child he’d met on Roblox.

These predators weren’t lurking outside the world’s biggest virtual playground. They were inside, hanging from the jungle gym, using Robux to lure kids into sending photographs or developing relationships that moved to other platforms and, eventually, offline.

The Spawnism Cult

If you think it can’t get worse, I’m sorry to tell you it does.

In July 2025, an online cult called “Spawnism” emerged within the Roblox community. It centred around a fictional deity from a popular Roblox game called Forsaken. What started as game lore became something far more sinister when predators realised they could use the “cult” framework to target vulnerable children.

These predators would recruit children into Spawnism, then manipulate them into carving the cult’s symbol into their own skin, performing degrading acts on camera, and committing severe self-harm. All of this coordination happened on Discord, with the initial recruitment and grooming happening through Roblox.

Children were literally mutilating themselves because predators on a gaming platform had convinced them it was part of belonging to something.

This is what happens when platforms prioritise growth over safety. This is what happens when moderation is gutted in favour of AI efficiency. This is what happens when corporate executives know there’s a problem and choose profits instead.

82 Million Users, 3,000 Moderators

The platform employs about 3,000 moderators for 82 million daily active users. That’s one moderator for every 27,333 users. And that’s assuming all 3,000 are working simultaneously, which obviously they’re not.

Former moderators have spoken to journalists about what it’s actually like trying to keep children safe on Roblox. One told Bloomberg that her team received hundreds of escalated reports involving child safety every single day. Hundreds. Daily. And this was just the reports that made it through the initial AI screening to reach human eyes.

Current and former trust and safety workers consistently say the same thing: user growth takes priority over child safety at Roblox.

Proposals for stronger safety settings that align with regulations like the UK’s Age Appropriate Design Code are shelved because leadership doesn’t want anything that might reduce engagement metrics.

Roblox publicly supported the California Age Appropriate Design Code. They released statements about being “proud to be one of the first companies” to back similar legislation. But behind closed doors, according to their own former employees, they were rejecting the very safety measures those laws were designed to ensure.

It’s corporate theatre. Safety-washing. Telling parents and regulators what they want to hear whilst doing the absolute minimum required, and sometimes not even that.

The maths alone is absolutely insane. With 82 million daily users and over 50,000 chat messages processed every second, how can 3,000 moderators possibly keep up? They can’t. And Roblox knows they can’t. But admitting that would mean spending more money on moderation, which would hurt their bottom line.

My Assessment After My Research

I started this investigation because 5 year olds at my daughter’s school were playing a violent slapping game they’d seen on Roblox. I thought maybe the platform needed better age ratings, better parental controls, better education for parents about settings.

What I found instead was a company that has enabled the systematic exploitation of children at scale, that has repeatedly chosen profits over safety, and that continues to operate with minimal consequences despite overwhelming evidence of harm.

So here’s my verdict, and I’m not going to soften it...

If your child is under seven, they shouldn’t be anywhere near Roblox or any online gaming platform. Not with parental controls. Not with supervision. Not at all. They should be playing with physical toys, playing outside, playing on a simple handheld device like a Game Boy if you want them to have screen time. But not online. Not where their every word can be recorded, monitored, and potentially seen by strangers.

The infant school children imitating that slapping game shouldn’t even know what Roblox is. And the fact that they do tells you everything about how normalised it’s become to hand young children devices with unfettered internet access before they can even read properly.

For children aged seven to twelve, my honest recommendation is to avoid Roblox entirely if possible. I know that’s difficult when it feels like every child at school uses it. I know the social pressure is real. But so is the risk. If you decide to allow it anyway, you need to understand that you cannot rely on Roblox’s safety features, you cannot trust their moderation, and you cannot assume that the parental controls will actually work.

That means you’re taking on the full burden of safety yourself. You need to disable all chat features... not just limit them, disable them completely. You need to turn off private messaging. You need to restrict your child to games from verified creators only, and even then, you need to regularly check their play history. You need to disable all in-app purchases because predators use Robux as bait. You need to have ongoing conversations with your child about what grooming looks like, what red flags to watch for, and why they should tell you immediately if anyone asks them personal questions or makes them uncomfortable.

And even with all of that, you’re still trusting a platform that has demonstrated, repeatedly and definitively, that it values user growth more than child safety.

For teenagers thirteen and up, Roblox can potentially be used more safely, but only with proper education about online dangers, ongoing monitoring of their activity, and clear boundaries about what’s acceptable. Teenagers are more capable of recognising manipulation tactics, but they’re also at a vulnerable age for body image issues, peer pressure, and risk-taking behaviour. The Spawnism cult specifically targeted vulnerable teens. The grooming that led to Ethan Dallas’s suicide started when he was twelve and continued until he was fifteen.

Age doesn’t make them immune. It just changes the nature of the risks.

What can you do as a parent?

I know this has been heavy. I know it’s easy to feel overwhelmed and paralysed by information this disturbing. But there are practical steps you can take right now, tonight, before your child logs into Roblox again.

First, and this is non-negotiable, log into your child’s Roblox account and check their message history. Don’t give anyone time to delete anything. Just check. Look at who they’ve been talking to. Look at what’s being said. If you find anything concerning, screenshot everything before you do anything else. You might need that evidence.

While you’re in the account, go to Settings and then Privacy. The default settings on Roblox are dangerous. They’re designed to maximise engagement, not safety. You need to manually restrict everything. Turn off the ability for anyone to send messages. Turn off the ability to join games with anyone except friends you’ve specifically approved. Disable voice chat entirely. Set the account to private so strangers can’t see your child’s activity.

Then go to Parental Controls and set up a PIN that your child doesn’t know. This prevents them from changing any settings without you. Yes, this might feel like you’re being controlling. But after what I’ve shown you about what’s happening on this platform, I think we’re well past worrying about being the “cool parent.”

Check the “Continue Playing” history to see which games your child has been accessing. Google those game names separately to see if there have been any reports of inappropriate content. If you find anything concerning, block those specific games through the parental controls.

Have a conversation with your child that’s honest without being terrifying. You don’t need to tell a seven-year-old about child pornography trading groups. But you can explain that some adults pretend to be children online to trick kids into sharing personal information or pictures. You can establish the rule that if anyone online asks them questions about where they live, what school they go to, or asks them to keep secrets from you, they need to tell you immediately and they won’t be in trouble.

For older children and teenagers, you can be more direct. Show them news articles about Ethan Dallas if you think they’re mature enough to handle it. Explain that predators specifically target gaming platforms because children naturally trust people they play games with. Make sure they understand that anyone can claim to be any age online, that voice-altering technology exists, and that even someone who sounds like a child might not be one.

And then, and this is the part that’s hardest but most important, you need to make an ongoing commitment to monitoring your child’s online activity. Not once. Not this week because you read an alarming article. Ongoing. Weekly at minimum. Check those message logs. Check the play history. Check friend requests and who they’re connected to. Make it routine, make it non-negotiable, make it clear that online privacy isn’t absolute when you’re a child using platforms that have proven they can’t be trusted.

The Alternatives Worth Considering

I know some of you are thinking, “This all sounds extreme. Surely there are safer ways for children to play online games.”

There are, but they require parents to be much more involved and much more selective.

Minecraft, when played on private servers with only people you know in real life, is significantly safer. Yes, it’s still online. Yes, there are still risks. But the difference is that you can completely control who your child interacts with. Set up a family server, or a server just for their real-world friends whose parents you actually know. No public servers. No joining random worlds. Just a closed environment where you know every single person they might encounter.

For younger children under nine or ten, I’d honestly recommend sticking to completely offline gaming. A Nintendo Switch has hundreds of age-appropriate games that don’t require any internet connection. Old-fashioned Game Boys or DS systems can be bought second-hand cheaply and have massive libraries of games designed for children. Board games. Card games. Outside games. I know that sounds impossibly old-fashioned in 2025, but you know what those options don’t have? Groups of adults trading child pornography. Predators using voice-altering software. Cults recruiting children to self-harm.

Sometimes old-fashioned is actually just... safe.

The Conversation That Matters Most

After I saw that school app post about the slapping game, after I started researching and found all of this, I had a conversation with my daughter that I’d been putting off. I’d been telling myself she was too young, that I didn’t want to scare her, that surely the parental controls would be enough.

But here’s what I realised... the discomfort I felt having that conversation was nothing compared to the harm that could happen if I didn’t have it. My temporary discomfort about discussing online safety was insignificant next to the very real danger of remaining silent.

So we talked. Age-appropriately, without graphic details, but honestly. We talked about how not everyone online is who they claim to be. We talked about red flags. We talked about the rule that she can tell me anything, show me anything, ask me anything about her online interactions and she will never, ever be in trouble for it.

And then we talked about why she doesn’t need to be on platforms like Roblox right now, why we’re making different choices than some of her friends’ families make. Not because we don’t trust her, but because we don’t trust the companies running these platforms to keep her safe.

That conversation was uncomfortable. She didn’t love hearing that she couldn’t do everything her friends do. But it was necessary, and it opened a dialogue that we’ll continue having as she gets older and as the online landscape continues to evolve.

Have you had a conversation like this already with your children? If not, maybe it is time.

If this article has raised concerns about your child’s online activity, trust your instincts. Check their accounts tonight. Have the difficult conversations. And if you discover something concerning, document everything and report it to police immediately. In the UK, you can report concerns to the Child Exploitation and Online Protection Centre (CEOP) at www.ceop.police.uk. In the US, report to the National Center for Missing and Exploited Children at CyberTipline.org.

The proof came in the most mundane way imaginable.

Danielle was having a perfectly normal evening in her Portland home, chatting with her husband about hardwood floors, when her phone rang with news that would eventually trigger one of the largest children's privacy fines in history.

"Unplug your Alexa devices right now," urged the woman on the other end, her husband's colleague calling from Seattle. "You're being hacked."

But Danielle and her husband weren't being hacked.

They were being surveilled.

The colleague had received audio files of their private conversation. Their Amazon Echo had recorded their discussion without any wake word being spoken, then helpfully sent that recording to a random contact in their phone book.

When Danielle's husband initially didn't believe it, the Seattle colleague delivered the proof that changed everything:

"You sat there talking about hardwood floors."

That awkward phone call in 2018 became the thread that unraveled Amazon's entire surveillance operation.

The £20 Million Truth Amazon Hoped You'd Never Find Out

What started as one family's bizarre privacy invasion became evidence of something much bigger and more sinister. When Amazon's engineers investigated, they confirmed exactly what had happened: "They said 'our engineers went through your logs, and they saw exactly what you told us, they saw exactly what you said happened, and we're sorry.' He apologised like 15 times in a matter of 30 minutes".

Those apologies felt pretty hollow five years later when the Federal Trade Commission and Department of Justice slapped Amazon with a £20 million fine for systematic violations of children's privacy laws, one of the largest penalties under America's child protection regulations.

The government investigation revealed what Amazon had been quietly doing: this wasn't a one-off glitch but a systematic operation affecting over 800,000 children under 13, with Amazon "keeping children's recordings indefinitely, and flouting parents' deletion requests".

Your children's innocent questions, bedtime conversations, and family moments were being kept forever and used to train algorithms designed to better understand and influence them.

The Surveillance Operation Hiding in Your Kitchen

Federal investigators discovered that Amazon had built something rather clever: a surveillance network disguised as helpful household gadgets.

Here's what they were actually up to:

Hoarding children's private conversations: Amazon kept children's recordings indefinitely by default, breaking rules that say such recordings should only be kept as long as "reasonably necessary"

Ignoring parents who tried to protect their kids: Even when parents specifically asked Amazon to delete their children's data, the company failed to remove transcripts from all their databases

Turning childhood innocence into profit: Amazon used children's unlawfully kept voice recordings to train their Alexa system, because "children's speech patterns and accents differ from those of adults," giving them "a valuable database for training" that helped "its bottom line at the expense of children's privacy"

Lying to parents about what they were doing: Amazon "prominently and repeatedly assured parents that they could delete voice recordings," but "failed to follow through on these promises when it kept some of this information for years"

As the government investigators put it: "Amazon's history of misleading parents, keeping children's recordings indefinitely, and flouting parents' deletion requests violated COPPA and sacrificed privacy for profits".

Not exactly the helpful family assistant they advertised.

The Legal Battle That's Still Going

The government fine was just the start. Multiple court cases are ongoing, with judges allowing families to sue Amazon as a group for allegedly recording and storing conversations without permission.

Comments from affected families show just how widespread this surveillance was:

"I have multiple Alexa devices. The one in my kitchen randomly responds to conversations I am having with other people."

"Does your Alexa do things without being asked?"

These aren't software bugs, they're the inevitable result of a system designed to hoover up as much family data as possible.

What Amazon Never Told the Portland Family

When Amazon's engineer rang Danielle to apologise, the company claimed her family's privacy breach was just an "unlikely string of events." They didn't mention the systematic data harvesting affecting hundreds of thousands of other families.

The Portland family did what any reasonable people would do: they asked Amazon for a refund for devices that had violated their privacy.

Amazon said no.

The company offered to switch off some features but "representatives have been unwilling" to give them their money back.

Think about that for a moment:

Amazon was happy to pay £20 million to government regulators but wouldn't refund one family whose privacy violation helped expose their entire operation.

That tells you everything about their priorities.

The Surveillance You Welcomed Into Your Home

When you bought that smart speaker, you probably thought you were getting a helpful gadget that would play music and answer questions. What you actually installed was corporate surveillance equipment that runs 24/7, building detailed profiles of your family's most private moments.

Every smart speaker has multiple microphones designed to capture audio from entire rooms. These devices have to constantly listen to everything to detect their wake words, which means they're processing conversations, arguments, phone calls, and intimate family chats all the time.

For children, this surveillance is particularly invasive. Kids naturally treat Alexa like a friend, asking embarrassing questions, sharing playground drama, and repeating things they've overheard. All of this became permanent data used to train systems designed to understand and influence them for life.

The Real Cost of "Free" Convenience

Amazon's total privacy violation fines exceeded £24 million when you include both Alexa and Ring camera breaches, which suggests these weren't isolated mistakes but part of a business model that prioritised data collection over family privacy.

The maths is pretty simple: Amazon found your family's private conversations so valuable that they were willing to pay massive government fines rather than stop recording them.

Your children's speech patterns, interests, family dynamics, and innocent questions became corporate property worth more than the devices you bought.

What You Can Do Right Now

If you've got smart speakers around children, here's what you should do immediately:

Check what they've recorded: Log into your Amazon account and look at the stored recordings.

Understand that "delete" is a lie: Amazon's delete functions don't actually remove everything. Transcripts and other data stay in their systems forever.

Move the microphones away from private spaces: Get these devices out of bedrooms, bathrooms, and anywhere you have sensitive family conversations.

Turn off risky features: Disable purchasing, messaging, and calling functions that could be misused.

Keep records: If you find unauthorised recordings, especially of children, save everything. The ongoing court cases might mean you can get compensation.

Consider going old school: Normal Bluetooth speakers do the same job without the corporate espionage.

The Lesson That Cost Amazon £20 Million

The Portland family's story teaches us something important: when companies promise to protect your privacy, check what they're actually doing, not what they're saying.

Amazon's engineers confirmed they'd violated this family's privacy, apologised profusely, but still refused basic compensation. Meanwhile, they were harvesting identical data from hundreds of thousands of other families, building detailed profiles of children's development, family relationships, and private moments.

Our kids deserve better. They deserve homes where they can ask silly questions without corporate eavesdropping, where their innocent conversations stay private, and where their childhood isn't converted into algorithmic training data.

The Portland family trusted Amazon with their privacy and learned what that trust was really worth. Amazon trusted that families wouldn't discover their systematic surveillance and paid £20 million when that gamble backfired.

Next week: I'll show you what to use instead of corporate surveillance devices—tools that actually serve families rather than harvest their data.

Have you looked at your smart speaker's recording history yet? What unauthorised recordings did you discover? Your experience might help other families protect themselves or join the legal fight for compensation.