TikTok age bans are coming to Europe. Will they work?

On 1 September, a twelve-year-old in Lyon is supposed to lose her TikTok account. Not because a parent took the phone away. Because the law said so.

France’s parliament adopted the text on 21 July 2026, both chambers on the same day, making France the first country in the EU to put a flat ban on under-15s using social media into national law. According to LCP, the Assemblée nationale’s own channel, new accounts fall under the rule from 1 September and existing accounts belonging to under-15s have until 1 January 2027 before they are closed.

Whether that first date holds is not settled. The Sénat’s record of the bill notes that more than sixty deputies referred it to the Conseil constitutionnel on 23 July, and a second group did the same the following day. Until the Council rules, 1 September is a plan, not a fact.

The direction, though, is not in doubt. Which makes this a good moment to ask the question that gets lost in the shouting: when a country bans children from social media, what actually changes for the children?

Europe has stopped arguing about whether, and started arguing about how

The French law did not come out of nowhere. In November 2025 the European Parliament passed a resolution calling for an EU-wide digital minimum age of 16 for social media, video-sharing platforms and AI companions, with 13 as an absolute floor and parental consent as the only route in between. It carried by 483 votes to 92, with 86 abstentions. It is not binding on anyone. It was a very loud signal.

Since then, as Euronews reported in April, Denmark has struck a political deal on an under-15 ban, Greece is set to enforce its own from 2027, and Spain has been weighing 16. The European Commission has been building the missing piece: an open-source age verification app, meant to slot into national digital wallets, that uses zero-knowledge proofs so a platform learns you are old enough without learning who you are. Commission President Ursula von der Leyen said in April it was technically ready.

So the political question is close to settled. What is not settled is the bit that touches your child: who does the checking, and how.

The gate is an AI making a guess

On 16 January 2026, TikTok announced it was rolling out stronger underage-detection technology across the European Economic Area, Switzerland and the UK. The system reads profile information and on-platform activity, things like a profile image or a post mentioning an upcoming birthday, and predicts whether the person behind the account falls in a given age band. If the prediction clashes with the birthday given at sign-up, a specialist moderator looks at the account and decides.

TikTok says its moderators are told to err on the side of caution and remove the account where there is doubt. It also says the company already removes around 6 million underage accounts globally every single month, and that the European pilot led to the removal of thousands of additional underage accounts.

If the guess goes against a teenager who is genuinely old enough, the way back in is an appeal, and the appeal is a facial age estimation from Yoti, a credit card, or a government ID. We wrote about that trade when Discord rolled out its own checks.

Britain is a year ahead, and the numbers are humbling

The UK started requiring age checks under the Online Safety Act in July 2025, which means Britain has something the rest of Europe does not: evidence. Ofcom published its statutory report on 16 July 2026, and it is the most useful document in this whole story.

The scale-up was real. Over 69 million age checks were completed between July and December 2025 across a sample of 32 services, a 23-fold increase on the previous six months, and Ofcom says the true UK total is likely to be materially higher. Among 8 to 17-year-olds who recalled being asked to prove their age, 43% met a genuinely strong check in January 2026, up from 25% the previous July. Facial age estimation was the method children remembered most.

Then the awkward part. Ofcom’s tracker found that 73% of children aged 11 to 17 recalled seeing at least one piece of harmful content in the four weeks before being surveyed, against 70% before the duties came into force. The regulator’s own reading is that there has been little change. Meanwhile its March 2026 research found a quarter of 11 to 17-year-olds had used a VPN in the previous six months, rising to 31% of 16 and 17-year-olds.

And Ofcom is blunt about the exact technique the big platforms lean on. It says its analysis “reinforces our long-standing, serious doubts about the efficacy of some age inference models”, the kind that read a user’s behaviour to guess whether they are a child, and that in some cases those systems “may have failed to correctly identify large numbers of children on their platform”.

That is the regulator with the most operating experience in Europe, saying the gate went up fast and the view from the other side barely moved.

The read from Singapore

We are not spectators here. IMDA’s Code of Practice for Online Safety for App Distribution Services took effect on 31 March 2025 and requires designated app stores to work out which users are under 18 and stop them downloading the highest age-rated apps, with facial estimation as one permitted method. We looked at what that means in practice in our piece on app store age checks. Singapore has not banned under-15s from social media, and nothing in the French law reaches a phone in Punggol. But it is the same machinery, and it is already switched on.

The most honest line in any of this came from our own ministry. Asked in Parliament about these measures, MDDI said that “regulations alone are unlikely to prevent a parent or any adult from helping a child circumvent the age assurance measures.”

Our view: the gates are worth having. A default that makes the wrong thing harder is not nothing, especially for a nine-year-old. But a law with a start date is a floor, not a finish, and treating 1 September as the day the problem got solved is how a family ends up surprised in January.

So, this week: ask your child when a site last asked them to prove their age, and what they did next. The answer tells you more about where they actually are than any policy document will.