When I published part nine last month, I called it The experiment goes live. I should have waited for the projectionist. On 14 August, eighteen days before the first accounts were due to be switched off, France's Constitutional Council struck the whole thing down (France 24, PPC Land). The experiment will not go live in September. It may not go live at all in the form we were promised.

What the court actually did

The decision is n° 2026-911 DC, handed down after two groups of sixty deputies referred the law to the Council — the ordinary way a French statute is tested before it takes effect. The Council did not quibble at the edges. It censured Article 1 outright: the entire under-15 ban, the two-stage rollout, the universal verification — the machine I spent part nine describing.

Read the reasoning and you may feel, as I did, an odd sense of having heard it before. The Council gave three grounds. The first was proportionality: a restriction on freedom of expression and communication, protected by Article 11 of the 1789 Declaration, must be necessary, adapted and proportionate, and a blanket bar on every under-15 across every service — from the obvious platforms down to online games and messaging tools — was none of those; the risks it invoked were, in the Council's phrase, not established across the sweep of things it covered. The second was the absence of any individual assessment: no room for a parent to judge, no weighing of the child's age or the particular service, one rule applied to a fourteen-year-old, an eight-year-old and a chess app alike. And the third — the sentence I want to sit with — was that the age-verification requirement came with no guarantees: the law demanded that everyone prove their age but did not determine the conditions and limits of how, and so failed to protect the right to private life under Article 2 of the same Declaration (communiqué, JURIST).

The vindication I did not want to collect this way

I have said in nearly every instalment that I would rather France proved me wrong than right. It has instead been a French court that proved me right, and I have to be honest about how that feels, because it is not straightforwardly a victory lap.

The thing I called the cost nobody puts on the invoice — that to check everyone's age you must collect everyone's identity, and that mandating this without ironclad limits is a privacy harm in its own right — is, more or less, the thing the Council struck the law down for. The freedom-of-expression point is theirs, not mine. But the privacy point is one this series has pressed since part six. The honeypot I described — the permanent national store of the exact documents that ruin lives when they leak — will not be built this September, because the country's highest constitutional authority looked at the same design I did and decided the law had not troubled to protect the people it would enrol.

I will take that. But I want to be precise about what it is, and what it is not.

What this reprieve is not

It is not a change of mind about the goal. The Council was at pains to say the opposite: protecting children and public order are legitimate constitutional aims that can justify restricting access. It struck down this law's method, not its ambition. Within hours the Élysée had asked the prime minister, Sébastien Lecornu, to prepare a fresh text — one that is, in the government's word, legally sound, and mindful of both the ruling and the European framework — with the redraft aimed at spring 2027 (The Local, CNews).

Think about what curing this particular defect actually requires. The Council did not say collect less identity. It said the law had failed to determine the conditions and limits of verification. The obvious way to satisfy that is to specify the machinery in far more detail — to write the honeypot into the statute properly, with named methods and named limits, rather than leaving it to a regulator to improvise. A legally sound version of this law could just as easily be a more elaborate verification regime as a lighter one. The reprieve buys time; it does not settle the argument, and the next draft may be built more carefully around exactly the collection I warned about.

The part that survived — and why it is the interesting one

One piece of the wider package did not fall, because it was never referred: the ban on smartphones in lycées, which still begins on 1 September (The Next Web). I note this without irony, because it is close to the thing I have argued for all along. A phone that is off in a classroom is a device-level control, enforced where the child and the device actually are, by adults in the room — not a national identity checkpoint bolted to the front door of the internet. France has ended up, for now, with the measure that acts on the device and without the one that acts on the destination. That is very nearly the ordering part seven argued for, arrived at by the least likely route imaginable: a court striking out the destination law while the timetable quietly leaves the device rule standing.

The board has not really changed

Step back to the whole series and the direction of travel is dented, not reversed. Australia's own numbers, which I logged in part nine, showed a ban moving account ownership among under-16s by about ten points and overall access barely at all — the modest return this series predicted. Britain's under-16 ban and switch-off curfew are still drafted for next spring. The European Commission's Ursula von der Leyen is still pressing a Europe-wide social media delay for children. France has simply demonstrated something the others should read closely: that you can lose this in the courtroom as easily as in the schoolyard, and that the privacy shortcut is where a careful judge reaches first.

The enforcement inversion I keep returning to has a courtroom cousin, and France has just met it. Build the law in a hurry, hang the gate in the middle of the field, skip the guarantees — and the fastest circumvention of all turns out to be a constitutional one.

What changes for you

Nothing, which is rather the point. The advice at the close of every instalment stands, and this ruling does not move a word of it. When you are asked to prove your age online — and across Europe you increasingly will be, whatever becomes of any single law — treat your identity documents as the precious things they are, and prefer the providers that check and delete over the ones that check and keep. And keep doing the unglamorous, device-level work that has outperformed every statute in this series: the controls that travel with the child, and the conversation about why they matter that travels further still.

My prediction is still on the record, in part six, and this changes only its timing. The number I said I would watch most closely — how long a national identity store holds before something leaks out of it — has been given a stay of execution, because in France that store will not be built this autumn. I would far rather report a reprieve than a breach. But a reprieve is what this is: the redraft is already commissioned, the goal is unchanged, and the honeypot has been postponed, not cancelled. I will be here for the spring text, reading the guarantees line by line — because this time a court has told the government exactly which page to get right, and I want to see whether getting it right means collecting less of you, or simply collecting it more thoroughly.

This is part ten of Regulating the Teen Internet. Part nine, The experiment goes live, described the law this ruling has now struck down; the practical, do-it-this-weekend guide for parents is part eight, and it stands whatever any parliament or court does next.