France was about three weeks away from becoming the first country in Europe to lock everyone under 15 out of social media. Parliament had passed the bill in July with a comfortable majority. The rules were due to take effect in September. Platforms had started working out what compliance would look like. Then on Friday the Constitutional Council read the law properly and threw it out.
The decision is a serious blow to Emmanuel Macron, for whom this was shaping up to be one of the last significant legislative wins before his second term ends next year. It is also, and this part is getting lost in the headlines, not a ruling that children have a constitutional right to TikTok. The court did not say the goal was wrong. It said the law was built badly, in two specific ways that anyone drafting similar legislation elsewhere should read carefully.
What the Court Actually Objected To
The Council’s reasoning runs along two tracks, and they compound each other. The first is scope. The bill applied to online platforms that let users connect and communicate with one another, which sounds like a definition of social media until you notice how many things it describes. Collaboration software. Ordinary messaging apps. Online games that happen to include a chat window or a friends list.
The court’s objection was that a great many of those services have never been shown to pose the risks the law was written to address. Sweeping them in anyway means restricting minors’ expression and communication in places where there is no demonstrated harm to justify it, and that is what tipped the measure into disproportionate.
The second track is privacy. A ban by age only functions if you can establish age, and establishing age at national scale means every user proving how old they are. The law required the checks without building adequate protection around the data those checks generate. The Council found that insufficient, which matters because it is the harder of the two problems to fix.
| The objection | Why it was fatal | How hard is it to fix? |
|---|---|---|
| Definition too broad | Covered games, collaboration tools and messaging apps with no proven risk to minors | Manageable. Narrow the definition and name the services |
| No privacy safeguards on age checks | Verification would collect identity data from every user with no protective framework | Hard. This is an unsolved problem almost everywhere |
| Disproportionate limit on expression | The two problems above together outweighed the protective benefit | Follows from fixing the first two |
The Age Verification Trap
This is the part worth dwelling on, because it is the reason similar laws keep stalling in different countries for reasons that look unrelated but are not.
To keep a 14 year old off a platform, you have to know that a given user is 14. There is no way to know that about the minority without checking the majority. So a rule aimed at children turns into an identity check applied to the entire adult population of a country, and the platform, or a vendor it hires, ends up holding a pile of government ID scans or biometric estimates belonging to tens of millions of people who were never the point of the law.
That pile is a target. It is also a permanent change to what using the internet anonymously means in that jurisdiction. The Council did not rule that age verification is inherently unconstitutional, which is an important distinction. It ruled that this law imposed it without building the protections that would make it acceptable.
France Was Following Australia, Not Leading
The French bill was explicitly modeled on an experiment already running. Australia’s under-16 ban, the first of its kind anywhere, took effect in December and cut off access to platforms including Facebook, Snapchat, TikTok and YouTube. France’s July vote would have made it the first European country to follow.
The comparison is instructive because the two countries approached the same idea from different legal starting points.
| Australia | France | |
|---|---|---|
| Age threshold | Under 16 | Under 15 |
| Status | In force since December | Struck down before taking effect |
| Scope | Named platforms | Any service allowing users to connect and communicate |
| Constitutional free speech guarantee | No explicit general right in the constitution | Yes, and the Council applied it |
That last row does a lot of the work. Australia’s law survived partly because there was no equivalent constitutional hook to hang a challenge on. France has one, the Council used it, and the broader the French definition got, the more it exposed itself.
What Happens Next
Macron has asked the government, under Prime Minister Sébastien Lecornu, to produce a redrafted version. The stated target is to have something in force before spring 2027, which is a realistic timeline for a rewrite and a tight one for a rewrite that has to solve the privacy problem.
The likely shape of version two is not hard to guess. Expect a narrower definition that names categories of service or specific platform characteristics rather than describing anything social. Expect an age verification framework built into the law itself, with data minimization rules, retention limits and probably a preference for device-level or third-party attestation over platforms holding identity documents directly.
What is less clear is whether a version narrow enough to survive review is still strong enough to do anything. A ban that covers four named apps and exempts everything else invites teenagers to simply move, which is the pattern every content restriction of the last twenty years has produced.
Why This Matters Outside France
Every legislature currently drafting a minors-and-platforms bill just received a free case study in what a court will accept and what it will not. The two failure points the Council identified, overbroad definitions and unprotected verification, are the two most common features of these bills worldwide, which means a lot of pending legislation has the same weaknesses sitting in it right now.
The United States is working through a parallel version of the same argument from a different angle, focused less on general social media access and more on specific product categories, which is why the Senate’s push to put AI chatbots on a leash for kids has privacy advocates genuinely split rather than uniformly opposed. The protective goal is widely shared. The mechanism is where everyone disagrees.
Elsewhere the pressure is arriving through the courts rather than the legislature. In China, a father suing two of the largest gaming companies over his son’s spending turned into a much bigger story than the sum involved, and the symbolic $1.50 lawsuit against Tencent and miHoYo was designed from the start to test where responsibility sits when a minor is on the other side of the screen.
The Bottom Line
Nothing changes for French teenagers this September. The ban that was weeks away is now a redraft with a 2027 target, and the government has to solve a problem that nobody has solved cleanly anywhere.
The interesting part is that the court did not tell France it cannot protect children online. It told France that a law which restricts a fifteen year old from a homework collaboration tool, and hands over the identity documents of forty million adults to do it, is not the way. That is a narrower ruling than the headlines suggest, and a more useful one. The next draft will be better because of it, assuming anyone reads the reasoning rather than the verdict.
