On July 21, 2026, the French National Assembly passed a bill banning social media use for those under 15 by a vote of 279 to 81. The Senate passed it the same day, with 243 in favor, 2 against, and 100 abstentions. Support crossed party lines. In the initial vote in January, the tally had been 130 to 21. Just two days after passage, on July 23, Prime Minister Lecornu referred part of the bill to the Constitutional Council—a move that revealed the government itself recognized the risk of unconstitutionality. Lawmakers from the left-wing party La France Insoumise also filed suit, arguing the bill was unconstitutional.
Three weeks later, on August 14, the Constitutional Council, in Decision No. 2026-911 DC, ruled Article 1 of the bill unconstitutional. The conclusion reached by the nine judges (including former Prime Minister Alain Juppé and former National Assembly President Richard Ferrand) was unambiguous: the ban was "neither adapted, necessary, nor proportionate" (ni adaptée, nécessaire et proportionnée) to the objective pursued.
What this ruling identified was not a procedural flaw. It is the first case in which a constitutional-level answer has been given to a structurally unavoidable problem for democratic states: the collision between protecting children and freedom of expression.
How Far Did the Definition of "Social Media" Extend the Reach of the Ban?
The Constitutional Council's decision rests on three points. The first is that the scope of the ban was too broad.
The bill defined the targets of the ban as "any online platform that enables users to connect with one another, communicate, share content, and discover other users or content." Under this definition, online encyclopedias, educational directories, and open-source development platforms were excluded, but collaborative content-sharing services, online communication apps, and even online games with cooperative-play features fell within the scope of the ban.
The Council pointed out that not all of these services necessarily pose risks to minors' health or safety. A design that blocks access uniformly, without individually assessing whether risk exists, was judged not to satisfy the principle of proportionality as a restriction on freedom of expression.
The second point concerns the exclusion of parental authority. The bill contained no mechanism whatsoever for parents or legal guardians to lift the ban, or to permit access to specific services, in the interest of the child. The Council cited the absence of provisions allowing for individualized risk assessment based on the child's age, maturity, family environment, and the nature of the service in question as one of the grounds for unconstitutionality.
Third, the legal safeguards for age verification were insufficient. Excluding those under 15 necessarily requires all users (including adults) to prove their age. However, the bill did not set out the conditions or limits of such proof, and thus failed to provide the necessary legal protection for the right to privacy (Article 2 of the 1789 Declaration of the Rights of Man).
| Issue | Bill's Design | Constitutional Council's Finding |
|---|---|---|
| Scope of the ban | All platforms enabling connection and communication between users | Uniformly blocks services with unproven risk. Violates the principle of proportionality |
| Relationship to parental authority | No mechanism for parental lifting of the ban or exceptions | Provisions allowing parents to make judgments based on the child's individual circumstances are necessary |
| Age verification | Requires age proof from all users | Conditions and limits unspecified. Lacks legal safeguards for the right to privacy |
The "Wall of Enforcement" Revealed by Australia
Another reason France's ruling of unconstitutionality has drawn attention lies in the struggles of the preceding case: Australia.
Australia enacted the Social Media Minimum Age Act in 2024 and, from December 2025, banned social media use for those under 16—the world's first such attempt. However, a follow-up survey conducted three months after implementation, between March and April 2026, found that more than 86% of those under 16 who had used the banned platforms continued to maintain access.
A study published in the BMJ (using a regression discontinuity design) reported no statistically significant discontinuity in social media usage rates around the age-16 threshold (P≥0.60). Daily social media use among 12- to 13-year-olds remained flat; among 14- to 15-year-olds it declined slightly from 78% to 69%; but among those 16 and older, it actually rose from 80% to 89%.
Workarounds were also varied. The proportion of adolescents using someone else's account rose from 9% to 29%; the creation of fake accounts rose from 15% to 19%; and the use of private browsers rose from 6% to 11%. VPN use remained a minority at 2% to 3%, but this indicates that age verification was so ineffective that a VPN wasn't even necessary. In fact, 66% reported having encountered age verification, and among these, the most common methods were self-declaration (24% to 39%) and uploading a selfie (13% to 27%).
Australia's eSafety Commissioner, while reporting the decline in account ownership from 52% to 42% as a "statistically significant decrease," also acknowledged that the majority still maintained access. The primary cause cited was insufficient implementation of age-verification measures on the platform side.
The Technical Puzzle of Age Verification
The age-verification requirement flagged as constitutionally problematic in the French bill also faces unresolved technical challenges.
A large-scale evaluation of facial age-estimation algorithms conducted by the U.S. National Institute of Standards and Technology (NIST) in 2024, using tests involving approximately 8.5 million subjects and roughly 11.5 million photographs, found that even the best algorithms had a mean absolute error (MAE) of three to five years for teenagers. Across all algorithms tested, the proportion of 13-year-old subjects whose age could be estimated within one year of their actual age remained below 35%.
A UK government survey report on age verification published in 2026 acknowledged that the accuracy of facial age estimation has improved in recent years (Yoti's algorithm achieved an MAE of about 2.0 years for ages 6 to 17), but noted that "cases remain where a 12-year-old is misclassified as 14, or a 16-year-old as 18." Accuracy biases by gender, skin color, and ethnicity have also not been resolved.
A technical assessment report from the Knight-Georgetown Institute points to an even more fundamental constraint: "All age-verification systems are vulnerable to circumvention, and it is technically impossible to build a system that completely prevents minors from accessing a service while not also blocking large numbers of adult users."
| Metric | Australia (before ban) | Australia (3 months after implementation) |
|---|---|---|
| Usage rate of banned platforms among under-16s | 85.9% | 81.5% or higher |
| Account ownership rate | 52.4% | 42.1% |
| Proportion who encountered age verification | N/A | 66% (self-declaration most common) |
| Daily social media usage rate (all ages) | ~60% | ~58% |
The Legislative Race Among European Countries and the Impact of France's Ruling
France's ruling of unconstitutionality has a direct impact on other European countries preparing similar legislation.
The UK government conducted a large-scale public consultation between March and May 2026 and has settled on a policy to implement a ban on social media use for those under 16 in spring 2027. Reportedly, 90% of parents support the ban, and two-thirds of young people also agreed. Following the Australian model, the ban would target Instagram, YouTube, TikTok, Snapchat, Facebook, and X. Ofcom is currently developing technical standards for age verification.
Norway's government has proposed an approach that grants access from January 1 of the year a child turns 16—a design in which children of the same school grade gain access rights simultaneously. Karianne Tung, Minister for Digitalisation and Public Governance, stated: "Rather than placing the responsibility on children, we place the responsibility on the platform companies."
Greece announced in April 2026 its plan to implement a ban on those under 15 effective January 1, 2027, and stated it would impose fines of up to 6% of a platform's global revenue on violating platforms, based on the EU's Digital Services Act (DSA).
The lesson these countries should draw from France's ruling is not about the legitimacy of a ban itself. The Constitutional Council acknowledged that the legislative purpose of protecting minors from online risks does serve "the realization of the child's best interests, a constitutional requirement," and it did not deny the legitimacy of the public interest involved. What was called into question was the design of the means. Narrowing the target based on the actual state of risk, incorporating a mechanism that respects parental judgment, and attaching legal safeguards to age verification—unless these three conditions are met, similar legislation risks hitting a constitutional wall in any democratic state.
Macron's Remaining Time and the Hurdles the Revised Bill Must Clear
Immediately following the ruling of unconstitutionality, the Élysée Palace announced that President Macron had instructed Prime Minister Lecornu to revise the bill, expressing intent to have it implemented by spring 2027. France's presidential election is scheduled for a first round of voting on April 18, 2027, and a runoff on May 2; Macron cannot run again due to the constitutional two-term limit. The social media ban was meant to be one of the few major legacies he could leave before his term ends.
For a revised bill to meet the standards set by the Constitutional Council, at least three design changes would be needed. First, limiting the scope of the ban to services with demonstrated risk, or introducing a risk-assessment mechanism for each service. Second, establishing provisions for parental lifting of the ban or conditional permission. Third, explicitly stipulating in the statutory text the conditions, limits, and data-protection legal framework for age verification.
However, the moment these revisions are made, the bill loses its original character as a "blanket ban." The closer it moves toward risk-based individualized regulation, the more it becomes regulation rather than a ban. The more broadly parental exceptions are recognized, the more its effectiveness is diluted. And the more strictly legal safeguards for age verification are demanded, the further the tension with privacy deepens.
As the Australian BMJ paper pointed out, it may take a decade for the full impact of a policy to become apparent. But lawmakers only have until the next election. What France's Constitutional Council demonstrated was not the legitimacy of the goal of protecting children, but the outline of the procedures that democratic states must observe in pursuing that goal. How much effective regulation can be built within this outline—no country yet has the answer.