On 21 July 2026, the French Parliament passed the law banning social media for minors under 15.
On 14 August 2026, it was struck down by the Constitutional Council for several reasons, among them the fact that it was a disproportionate infringement of freedom of expression and communication and of the right to privacy. Moreover, it took no account of minors’ age, family situation or degree of maturity, and did not allow parents to authorise access to certain services in their child’s interest.
On 26 August 2026, Meta was accused by 47 US states of deliberately introducing algorithms likely to create addictive behaviours in young children. It was accused of having shown what could be called “passive predation” by opposing all the safety measures that its own research teams strongly suggested. And it also showed “active predation” by using knowledge from neuroscience to create random-reinforcement learning loops in order to reduce users’ ability to resist the urge to return to its platforms. In a settlement intended to spare it from paying 200 billion dollars and going to prison, Zuckerberg agreed to pay 18 billion, undertaking to make changes to how its platforms work. They remain off-limits to under-13s, but several reforms are introduced: a limit of two hours a day on Instagram and Facebook for under-18s, blocking of access at night, interruption of notifications during school hours and screen-time warnings in case of uninterrupted use. Teenagers also have the option of choosing a feed not personalised by algorithms and of switching off automatic video chaining. These commitments are not trivial: they introduce, into the very design of the network, the possibility of stopping. But Zuckerberg offered no apology, and the economic model of the attention economy is not called into question.
Finally, on 17 September 2026, the European Commission adopted the EU KIDS regulation to strengthen children’s online safety across the Union1, with two flagship measures.
- A ban on access to social media before age 13, access under parental control from 13 to 15, then autonomy from 15. A harmonised threshold across Europe, where until now each country was cobbling together its own rule.
- A change of logic: it is no longer for children or parents to cope on their own; it is for the platform to prove that its product is safe and designed for the user’s age from the outset. An end to infinite scrolling, intrusive notifications and geolocation by default for the youngest, unless justified.
These concerns match exactly the 3-6-9-12 milestones, which have proposed since 2008 no screens before 3, no personal digital device before 6 (in 2008, this applied only to game consoles because tablets did not exist), no internet before 9 and accompanied internet between 9 and 12 completed, so 13, which meant “no smartphone before 13” and therefore no access to social media either.
Finally, the European text also targets chatbots and “AI companions”. Consumer conversational AIs will have to limit anything that creates emotional dependence in a minor, no longer remember their conversations, and switch off microphone and geolocation by default. The real task is no longer only social media; it is the whole relationship of children with machines that talk to them, and this also matches the concerns of the 3-6-9-12 association.
But this law will only be truly effective if it is accompanied by other measures.
Education first. This means that of children, but also that of parents. The Prime Minister’s announcement of one hour a week devoted to AI in Year 11 will not stop 10- or 11-year-olds from putting their questions to ChatGPT behind adults’ backs, at the risk of becoming dependent on it. This education must begin in primary school and cover both screens and AIs. Some US states are considering allocating part of the sums paid by Meta to prevention programmes. What a good example for Europe!
As for parents, a great many of them are asking to be helped in their decisions by clear, scientifically validated information. The American decision recognises parents’ educational role by allowing them to change the default settings. It is a way of encouraging family dialogue.
Creating alternatives to screens
It is not possible to forbid children from meeting their friends on social media without letting them meet in person. The State must strongly encourage local authorities, schools and regions to draw up agreements that give them free access to the collective play spaces in their area, starting with schools. And even, let us dream a little, to give everyone access to paid activities on a sliding scale, along the lines of that used for school canteens.
The design of services. The platforms’ business model is not called into question by the agreement reached in the United States, and this is where European law must take over, and not only for children. All services must be designed by default so as not to make users dependent, and their ability to choose must be built in at every level. In other words, platforms must stop being machines for capturing attention and become tools for connecting people. The possibility of personal settings must be there from the start of every account.
This is the aim of another European text (the Digital Fairness Act), due in the autumn, to tackle the addictive design of digital technology more broadly. Negotiations on combating online sexual abuse are continuing in parallel.
The issue is not screen time, it is mastery of attention. American justice and the European Commission converge on one essential point: what appeared in platform design to be a supposed technical inevitability actually depends on political decisions.
