A Blanket Ban Meets a Proportionality Test
On August 14, 2026, France's Conseil constitutionnel struck down Article 1 of the loi visant à protéger les mineurs des risques auxquels les expose l'utilisation des réseaux sociaux — the law that would have barred anyone under 15 from creating a social media account starting September 1, 2026, with existing accounts cut off from January 1, 2027. In Decision n° 2026-911 DC, the Council held the prohibition was "neither adapted, necessary, nor proportionate" to its stated goal, and separately found the age-verification regime it required lacked the legal safeguards the Constitution demands for private life.
The bill had cleared both chambers on July 21, 2026 after an eight-month legislative run that began with a National Assembly filing on November 18, 2025, following a parliamentary inquiry into TikTok's effects on minors. It was, briefly, poised to make France the first European country with a comprehensive under-15 social media ban.
The Case for the Ban, Stated Fairly
The underlying concern is not manufactured. Legislators built the bill on a documented record — an Assembly inquiry commission's findings on platform design and youth mental health — and the aim, protecting minors' well-being and development, is one the Constitutional Council itself accepted as legitimate and weighty enough to justify restricting a fundamental right. A single, bright-line age threshold is also the easiest rule to explain to parents, schools, and platforms alike, and easy rules are, in principle, easier to enforce evenly. If some subset of social platforms genuinely does drive compulsive use and harm among young teens, a categorical bar removes the burden from individual families of monitoring an addictive product family by family.
Why the Council Said No
But the Council's reasoning exposes exactly where that logic breaks down. Grounding its analysis in Article 11 of the 1789 Declaration of the Rights of Man, which the Council reads as protecting the freedom to access and speak on online public communication services, the judges found the ban's scope was defined by reference to the DSA's and DMA's broad platform categories rather than to actual, documented risk. As Covington's summary of the ruling notes, that definitional choice meant the law could have swept in collaborative and educational tools never shown to pose the harms the legislature was targeting — Wikipedia-style projects sat awkwardly beside Instagram in the same prohibited category. A law that cannot distinguish a moderated, education-oriented platform from an engagement-optimized feed is not narrowly tailored; it is a blunt instrument dressed as child protection.
The Council also faulted the bill for offering parents no mechanism to authorize access based on a child's maturity or family circumstances — removing the very layer of individualized judgment that would have made a categorical rule proportionate. And it went further on the mechanics: any enforceable under-15 ban necessarily requires verifying the age of every user, adult and minor alike, since a platform cannot know who to block without checking everyone. La Quadrature du Net had warned during the bill's passage that this amounts to "generalized identity control online," noting the text was silent on method but that officials had floated identity-document checks or integration with France Connect, the national digital-identity system. The Council agreed the legislature never specified the guarantees — data minimization, retention limits, purpose limitation — that would make such verification constitutionally tolerable.
The Right Failure Mode
This is, in the best sense, regulation working as designed: a legislature moved fast on a genuine problem, wrote an overbroad instrument, and a court forced a rewrite before it could inflict its worst side effects — mass identity verification for tens of millions of adult internet users, collateral damage to educational platforms, and no channel for parental judgment — on the entire population, all to reach a comparatively narrow slice of harmful platform design. A verification mandate that cannot tell a Wikipedia clone from a dopamine-optimized feed was never going to survive contact with either the market or a court; it is better that it be caught now than after every French adult had handed a government-linked identity check to log into a website.
What Spring 2027 Needs to Look Like
The Élysée has instructed the Prime Minister to draft replacement legislation aligned with the ruling and with EU law, targeting spring 2027 — before President Macron's term ends. Done well, that means grading restrictions to platform function and demonstrated risk rather than to a DSA-derived category label, building in a parental-authorization channel, and specifying privacy-preserving age-assurance methods — on-device estimation or zero-knowledge attestations, not centralized identity submission — with hard limits on data retention. Done poorly, it means the same categorical ban wrapped in cosmetic exceptions, headed for the same result. France has a real second chance to prove that child-safety regulation and an open, speech-protective internet are not actually in tension — but only if the next draft treats proportionality as a design constraint, not a drafting inconvenience to route around.