A tragedy, a nine-day bill, and a throttle switch
On April 23, 2026, Turkey's parliament adopted a law barring social media platforms from serving children under 15, mandating age verification, and ordering platforms with more than 10 million daily Turkish users to remove flagged content within one hour in urgent cases. It empowers the communications regulator, the BTK, to throttle bandwidth or block platforms that refuse to comply. The bill passed nine days after a 14-year-old killed nine students and a teacher at a middle school in Kahramanmaraş — the deadliest school shooting in Turkey's history — and was published in the Official Gazette in early May with a six-month compliance window before enforcement begins.
The case for acting is real — and is not the case for this law
The grief is real, and so is the concern. Investigators reported that the Kahramanmaraş attacker had immersed himself in online "incel" content and posted manifesto fragments on Discord. A parent who has watched a child radicalize through an algorithmic feed does not need a peer-reviewed citation to sense that something is wrong, and Turkey is hardly alone: Australia's under-16 ban and a wave of US state bills reflect a genuine, cross-partisan worry that platforms have offloaded the costs of engagement-maximizing design onto kids. A government that does nothing after ten funerals is not displaying principle; it is displaying paralysis.
But the strongest case for acting is not a case for this law. Two problems — one about evidence, one about power — make Turkey's response disproportionate to the harm it invokes.
The evidence does not carry the weight placed on it
The premise that social media use causes the harms now being legislated against is, at best, unsettled. The largest longitudinal studies find associations that are real but trivially small. In a 2019 analysis of UK adolescents published in PNAS, Orben, Dienlin and Przybylski concluded that "most effects are tiny—arguably trivial," with median within-person effects of roughly β = −0.05 and no statistical significance in more than half of their models. A 2020 review in Dialogues in Clinical Neuroscience landed in the same place: digital technology's effect on well-being is "very small—potentially too small to matter," and turns heavily on how a platform is used, with active social use often positive and passive scrolling negative.
As EFF argued this month, lawmakers keep selling correlation as causation, ignoring the selection effect that the psychologist Candice Odgers highlights — that struggling young people use these platforms more, not that the platforms make them struggle. None of this excuses harmful design. But a measure that severs an entire age cohort from mainstream platforms, and routes everyone else through identity checks, should rest on more than a moral panic compressed into a nine-day legislative sprint.
A child-safety law hands a censorship apparatus new leverage
The deeper problem is who holds the switch. The BTK's new throttling and blocking powers are not abstract — Turkey has used exactly these tools against speech before. Freedom House rates Turkey's internet "Not Free," scoring it 31 out of 100, and documents that authorities throttled or blocked X for roughly eight hours after the February 2023 earthquakes, while people were using it to call for rescue, and that more than 953,000 domains were blocked as of late 2023. The opposition CHP, which voted against the bill, warned of precisely this, pointing to platform restrictions during the 2025 İmamoğlu protests.
Layer a one-hour takedown mandate on top of that record and the incentives are obvious. When the penalty for leaving content up is bandwidth throttling or an outright block, and the deadline is sixty minutes, platforms will not adjudicate — they will delete. "Harmful to children" is an elastic category, and an elastic category enforced by a regulator with a documented appetite for political censorship becomes a fast lane for removing inconvenient speech, dressed as child protection.
Age verification is a tax on everyone's privacy
Confirming that a user is over 15 means confirming everyone's age — which in practice pushes platforms toward linking accounts to government identity, alongside Turkey's separate move to make all users verify their identity. That builds a record of who-said-what that no open society should want, and it does not even deliver the promised safety: as EFF and UNICEF both note, blanket bans tend to push the most vulnerable youth — LGBTQ+, rural, isolated teens — toward less-moderated corners of the internet where exploitation is more likely, not less. A measure justified by one boy's radicalization may make the next one harder to see.
A proportionate path exists
The CHP's framing — protect children "with rights, not bans" — is the right instinct, and it is not soft. The proportionate toolkit is well understood: enforceable transparency and risk-assessment duties, safety-by-default settings for minors, device- and OS-level parental controls that keep verification on the family's hardware rather than a state database, independent audits of recommender systems, and real investment in digital literacy. The EU's Digital Services Act pursues systemic-risk obligations without a blanket age cut-off or a regulator's throttle switch; that is the model worth borrowing.
Turkey's children deserve a serious response to a real tragedy. They are poorly served by one that rests on thin evidence, expands a proven censorship machine, and conscripts every adult's identity in the bargain. Safety and the open internet are not opposing goals — but you cannot secure the first by dismantling the second.