Macron Calls for AI Bans Where It Endangers Children

Emmanuel Macron spent the Paris AI safety meeting hosting a careful, consensus document. Then, in his own remarks, he said something the document does not. The French president called for artificial intelligence to be banned wherever it poses a risk to children's mental health, a sharper line than anything in the pledge the assembled companies had just signed.

Macron went further on process, too. According to reporting from the event, he said companies would accept the principle of a pre-audit and pre-certification, with a prior review of models before they are placed in children's hands. In plain terms, that would mean a product aimed at or accessible to minors could not ship until someone had checked it first.

A proposal, not a signed commitment

It is worth being precise about what this is. The Paris Call that the labs signed says nothing about children or minors. The protection of young users came from Macron's speech, which makes it a political proposal from a head of state rather than a binding undertaking by the firms in the room. The gap between the two documents is the whole story. The companies agreed to disclosure and shutdown mechanisms; they did not agree to let a regulator vet their products before launch.

How a ban would work in practice is also unresolved. AI services cross borders by default, and a French or European restriction does not reach a model hosted elsewhere. Pre-certification raises the same questions every safety-review regime does: who certifies, against what standard, and how fast, given that models now update in weeks.

Part of a wider turn toward minors

Macron is not speaking into a vacuum. Regulators have been circling the question of AI and young people for months. New York passed a law restricting AI companion chatbots for children, and an independent safety test this month found that every major chatbot failed a benchmark built around adolescent mental health. Against that backdrop, a call to keep unvetted models away from children reads less like a leap and more like a politician naming where the public is already uneasy.

Whether any of it becomes rule rather than rhetoric is the open question. Macron has floated the principle. Turning it into law, in France or across the European Union, is a longer road, and the companies have so far agreed only to the parts they wrote themselves.