The European Union is preparing a broad restriction on social media, AI chatbots, video-sharing platforms and online games for children under 15, according to a draft proposal reported by Reuters on Monday 14 September.
The planned EU Kids Act would create age bands rather than one universal switch. Under-threes would have no access; children aged three to 12 could use parent-controlled services designed for them; 13- and 14-year-olds could receive introductory accounts with strict parental controls; and users aged 15 and over could hold their own accounts.
The proposal is not law. Draft wording can change before publication, and any final act would require negotiation with EU member states and the European Parliament. Britain is outside the EU, although rules applied by global platforms can spill across borders and influence the UK debate.
More than a birthday gate
The draft reportedly covers Facebook, Instagram, TikTok, YouTube and ChatGPT, alongside gaming and other online services. Companies would need reliable age assurance, parental controls and simple ways to report harmful content. It would also target addictive design and charge platforms a supervisory fee.
That breadth reflects how children actually use the internet. Harm does not respect the categories in a policy document: a game contains chat, a video app becomes social media, and an AI companion can imitate emotional support while collecting intensely personal conversations.
Age verification is the engineering and privacy problem hiding beneath the political headline. A rule is only effective if platforms can distinguish a 13-year-old from a 31-year-old who has typed the wrong date deliberately. But asking everyone to upload identity documents can create a new store of valuable personal data and exclude people without conventional paperwork.
Possible methods include device-level age signals, facial age estimation, checks through trusted third parties and parental approval. Each brings error, accessibility and surveillance risks. The best system would disclose the minimum information necessary — for example, “over 15” — without handing a social network a passport copy and a fresh excuse to lose it.
What supporters and critics will argue
Supporters point to compulsive design, sexual and violent material, bullying, fraud and recommendation systems that can repeatedly amplify harmful content. AI companions add risks around manipulation, dependency and advice delivered with confidence but no professional duty.
Critics will question whether a blanket legal threshold respects teenagers’ access to information, peer support and creative communities. Vulnerable young people may rely on online groups unavailable locally. A badly designed ban could push them towards smaller services with weaker moderation or encourage a thriving market in borrowed adult accounts.
The age ladder partly recognises those trade-offs, but the details decide everything. “Parent controlled” must work for families without high technical confidence. Appeals are needed when an adult is wrongly classified as a child. Services must be clear about what data an age check collects, who receives it and when it is deleted.
Enforcement also has to reach the design choices that create compulsive use. Turning off infinite scroll, streak pressure, late-night notifications and personalised recommendation loops for children may achieve more than placing another easily ignored date-of-birth box at registration.
What it means in Britain
The UK’s Online Safety Act already places child-safety duties on regulated services and has driven age-assurance requirements for some content. An EU law would create a neighbouring large-market standard and increase pressure for clearer British rules on social media and AI companions.
OutOut will assess the published text when it exists. Until then, precise claims about bans, penalties or implementation dates should be treated as draft provisions, not instructions arriving in tomorrow’s school bag.
Follow the wider debate through OutOut’s Technology section.
The OutOut verdict
The EU has identified a real problem: online services know exactly how long a child watched a clip but can become mysteriously unsure whether the account belongs to a child.
A serious law must do more than move the “I am 13” button to “I am 15”. It should change product design, minimise identity data and make companies prove that protections work. Otherwise Europe will have constructed the continent’s most ambitious digital bouncer and left the fire exit open beside him.
