A Two-Month Tally, and a Revealing Ratio
On September 4, 2026, presidential communications director Burhanettin Duran announced that Turkish authorities had blocked access to 2,034 social media accounts over the preceding two months. The operation, coordinated by the Presidential Communications Directorate alongside the Chief Public Prosecutor's Office, the national police cybercrime unit, and the Justice and Interior Ministries, split the accounts into two buckets: 699 were blocked over alleged links to terrorist or criminal organizations — Duran singled out the Gülen movement (FETÖ) — and 1,335 were blocked for spreading "disinformation" or conducting what he called "psychological operations" against Turkey (Anadolu Agency).
That ratio is the story. Two-thirds of the accounts blocked had nothing to do with the terrorism allegations that dominate the government's framing — they were blocked for speech Ankara simply didn't like.
The Legal Mechanism, and Its Real-World Use
The blocking power comes from Article 8/A of Law No. 5651, Turkey's internet law, added in 2015. It lets the head of the telecoms regulator (BTK), acting on a request from the Presidency or a relevant ministry, order immediate blocking in "non-delayable" cases involving national security, public order, or crime prevention — implemented within four hours of notice, then submitted to a peace criminal court for approval within 24 hours, with the court required to rule within 48 hours or see the order automatically lapse (full English translation of Law 5651).
On paper, that is a judicial backstop. In practice, Freedom House's 2026 Turkey assessment found that of more than 311,000 web addresses blocked in 2024 — the highest annual total since Turkey started blocking sites in 2007, including over 5,700 news articles — 82% were imposed with no court order at all (Freedom House, Freedom in the World 2026: Turkey). The 24/48-hour review exists mostly as a formality that the administrative block has already outrun by the time anyone can contest it.
Who Actually Got Blocked
The named targets in the early-September wave illustrate the pattern. soL Haber, a left-leaning news outlet whose main account (924,000 followers) was blocked on July 3, saw its replacement account blocked again around September 3 — the second restriction in two months. Also blocked: Aile Dayanışma Ağı (Family Solidarity Network), a support group formed by relatives of politicians and municipal officials jailed in the sweeping investigations that followed Istanbul mayor Ekrem İmamoğlu's March 2025 arrest; İBB Tekzip, an account that rebuts allegations about the opposition-run Istanbul Metropolitan Municipality; İstanbul Gönüllüleri, a civic volunteer group dating to the city's 2019 election; and photographer-writer Fırat Erez (Stockholm Center for Freedom; İfade Özgürlüğü Derneği / EngelliWeb). İmamoğlu's own account, with roughly 9.7 million followers, was blocked back in April 2025.
None of this is a one-off. The Directorate has run near-identical announcements all year — 41 accounts in March, a cumulative 1,731 FETÖ-linked accounts by May — each pairing a genuine terrorism concern with a much larger, vaguer disinformation category that happens to catch government critics.
Steelmanning Ankara's Case
Turkey's security argument is not fabricated from nothing. The country has been the target of a violent coup attempt it blames on the Gülen network, faces an active PKK insurgency, and sits in a neighborhood — Syria, Iraq, the Caucasus — where online recruitment and coordination by armed groups is a documented problem. Democracies elsewhere, including EU member states applying the Digital Services Act, also compel rapid takedown of terrorist content with judicial review built in. A state that ignored genuine incitement to violence online would be failing a real duty. If Article 8/A were used only against the 699 accounts with concrete organizational links, this would be an unremarkable story about counterterrorism enforcement.
Where the Framing Breaks Down
It isn't used only that way. The other 1,335 accounts — the majority — were blocked under a standard so elastic ("psychological operations," "provocation") that it swept up a news outlet, a support network for jailed officials' families, and monitors of a mayor's own city hall. A 2022 disinformation law already criminalizes "spreading false information" with up to three years in prison; Article 8/A now adds a same-day administrative kill switch that operates almost entirely outside the judicial check the statute nominally requires. Turkey's Constitutional Court has previously found parts of the blocking regime disproportionate, but enforcement has outpaced any judicial correction.
The fix is not to abandon platform accountability — it is to make the two categories mean something. Genuine terrorist-content orders should carry public, reasoned justifications reviewable by an independent court before implementation, not after. The catch-all "disinformation" and "psychological operations" grounds, which now generate double the volume of the terrorism category, should require the same evidentiary threshold as any other prior restraint on political speech — or be repealed. A national security law that blocks a mayor's rebuttal account alongside a genuine militant recruitment channel isn't protecting national security; it's borrowing its language.