Why Age Bans Alone Won't Keep Children Safe Online

Why Age Bans Don't Work

The UK government is consulting on banning children under 16 from social media. Australia has already passed similar legislation. The political appeal is obvious: children are being harmed online, platforms haven't fixed it, so restrict access.

I understand the impulse. After years of voluntary commitments and incremental improvements, the harms continue. Parents are worried. High-profile cases make headlines. Something visible must be done.

But age bans, however well-intentioned, address only part of the problem. And focusing on them risks distracting from the harder work that actually determines whether children are safe online.

An age ban is an access control. It's a gate at the front door. The policy says: if you're under a certain age, you can't come in.

This is not nothing. Reducing the number of young children on platforms designed for adults has value. Age-inappropriate content, addictive design patterns, and social dynamics that harm adolescents are all real concerns. Limiting exposure to these has merit.

But an age ban is only as effective as the verification system that enforces it. And even perfect enforcement, which doesn't exist, only addresses who gets through the door. It says nothing about what happens once they're inside.

Consider what an age ban doesn't do:

It doesn't protect the 16-year-old who now qualifies for access but is targeted by an adult with harmful intentions. The legal line is arbitrary from a safeguarding perspective; vulnerability doesn't disappear on a birthday.

It doesn't address the determined 14-year-old who uses a parent's credentials, a VPN, or any of the other workarounds that young people have always found for age restrictions. The history of age-gated content online is not encouraging.

It doesn't change the fundamental architecture of platforms; the algorithmic amplification, the engagement-maximising design, the systems that surface risk and compound harm. These remain in place for everyone who gets through the gate.

It doesn't address the under-16s who are already on platforms, often with years of history, social connections, and established patterns of use. Enforcement against existing users raises its own complications.

The debate about age bans has absorbed enormous political and media attention. But it has also allowed us to avoid harder questions:

Why do platforms struggle to identify grooming patterns before harm escalates? Why does concerning behaviour compound for weeks or months before anyone intervenes? Why do safeguarding escalations stall between automated systems and human judgement? Why do platforms with extensive Trust & Safety teams still fail to protect vulnerable users?

These questions aren't about who gets access. They're about what happens to people who are already inside the system, and whether the architecture is designed to identify and respond to risk early enough to matter.

I'm not arguing against age verification or age-appropriate design. These have a place in a comprehensive approach to child safety online.

But they're access controls. They're gates. And gates are only one part of a building's safety infrastructure.

Safeguarding architecture is about what happens throughout the system: how risk is identified when it first appears, how concerning patterns are flagged and escalated, how decisions are made under pressure, where human judgement engages and with what authority.

A platform with strong age verification but weak safeguarding architecture will still fail children. A platform with imperfect age gates but robust systems for identifying and responding to risk early will protect more people.

If policymakers and platforms are serious about child safety, not just appearing serious, here's where I'd focus attention:

Require platforms to demonstrate how risk is identified early, not just how content is moderated after the fact. Examine escalation pathways: when something concerning is detected, what happens next? Scrutinise the AI-human handover: where does human judgement engage, and is that point appropriate to the risk?

These questions are harder than 'should under-16s be allowed on social media?' They're also more important.

Age bans may have a role. But they're not a substitute for the architectural work that actually determines whether systems are safe. The debate has given us a visible policy lever. The challenge is not letting it distract from the less visible work that matters more.