On August 14, 2026, France's Constitutional Council — the country's highest constitutional authority — struck down the central article of the law that would have banned social media access for all minors under 15. Emmanuel Macron has now tasked his prime minister, Sébastien Lecornu, with drafting a "legally sound" version before spring 2027.
This news has been read in two opposite directions. For some, it proves that banning is a mistake. For others, it's a procedural stumble that changes nothing. At Papitek, we don't buy either reading.
We do believe minors' access to social media needs to be regulated — particularly on Instagram and TikTok, the two platforms where most of the evidence on addictive design and harm to adolescent mental health is concentrated. What France just proved isn't that regulation is wrong — it's that there's a way to do it that doesn't survive two months in court. And Spain, which has its own law working its way through parliament, still has time not to repeat those mistakes.
Yes to regulation, no to looking the other way
Let's be clear from the start: the "let's wait and see" premise doesn't convince us. The design of these platforms — infinite scroll, variable notifications, algorithmic amplification of whatever hooks people most — isn't an accidental side effect, it's the foundation of their business model. A teenager isn't competing on equal footing against a product designed by entire teams of engineers to maximize time spent. That's exactly where someone needs to draw a line, because a family alone can't offset that imbalance of power.
That said, the interesting question has never been "regulate or not?" but what exactly gets regulated.
The problem isn't the minor, it's the design
The solution that actually attacks the root of the problem isn't deciding at what age a minor can join Instagram — it's forcing Instagram and TikTok to stop using the techniques that make it hard for anyone, not just a teenager, to put the phone down. Redesigning infinite scroll, limiting algorithmic amplification of content engineered to create dependency, eliminating variable-reward mechanics lifted straight from a slot machine: that would solve the problem at the source, for every age, without needing to pick an arbitrary age cutoff.
Interestingly, Spain's own bill points in that direction: it requires platforms to demonstrate to the CNMC (Spain's National Commission for Markets and Competition) that they don't use addictive designs and that they protect privacy before they can operate with minors in Spain. That's the good part of the idea. The problem is that, for now, that redesign depends on platforms wanting to do it, or on a regulator managing to prove it case by case — a slow process for a generation growing up with these apps right now, not five years from now.
In the meantime, patches don't cut it
And here's the nuance that matters most to us: the situation doesn't allow for half-measures. A penalty regime that fines a platform once the harm is proven — months or years after it happened — protects no one at the moment it matters. For a company with Meta's or ByteDance's revenue, a fine is an operating cost, not a real incentive to change the product. While the case is being litigated, regulated, and the violation proven, entire generations of teenagers stay exposed to the same design.
That's why an access restriction — imperfect, yes — makes sense as a bridge measure while the real redesign is forced through: it's the only thing that can be deployed now, not three years of paperwork from now. But precisely because it's an emergency measure, it can't afford to be badly designed. And that's exactly where France stumbled.
If you're going to restrict, do it without trampling basic rights
France's Constitutional Council didn't say "banning is wrong." What it said, in essence, is that this specific ban trampled basic rights in three ways:
Disproportion. A total cutoff, with no gradation or exceptions, restricting a minor's freedom of expression and communication.
Overreach. The law covered any social network, including platforms whose actual risk wasn't even proven.
Insufficient privacy in age verification. The mechanisms planned to check age offered no clear guarantees about what happened afterward to that sensitive data about the minor.
That third point, for us, weighs the most. If urgency forces us to legislate quickly with an access restriction, a minor's privacy can't be the variable sacrificed for speed. Verifying a teenager's age can't mean building a database with their ID document stored on some social network's server. That's not protecting the minor — it's trading one risk for another.
The mistakes Spain can still avoid
Spain's law, as currently drafted, shares at least two of the three French stumbles. It applies to "any social network" without distinguishing where there's real evidence of harm — the same overreach mistake. And it still doesn't specify any concrete age-verification mechanism — it defers that to a future regulation — which is exactly the piece that sank the French law on privacy grounds.
It does improve on France in one respect: for the 13-to-16 bracket, access requires both CNMC authorization and explicit parental consent — a margin for family decision-making that France never allowed for anyone. But that margin disappears for minors under 13, where the design reverts to the same exceptionless cutoff a court has already struck down in another country.
If Spain wants its law to actually work as an emergency bridge — and not as a patch a court rewrites within a year — it needs to resolve now, not in some future regulation, how to verify age without compromising sensitive data; grade the restriction by actual evidence of harm, not by blanket category; and demand, with the same insistence it demands the access restriction, the timeline and metrics for the platform redesign that's supposed to be the real end goal.
Banning without redesigning just postpones the problem. Redesigning without protecting, in the meantime, the people already inside is leaving them defenseless until that day comes. Spain doesn't have to get both calls wrong.