On May 26, 2026, Turkey's presidential Communications Directorate — working with the police cybercrime unit and prosecutors — blocked access to 93 social media accounts it accused of "provocative" posts aimed at "disrupting public order" and creating "social chaos." The blocks landed five days after the 36th Civil Chamber of the Ankara Regional Court of Justice, on May 21, annulled the main opposition Republican People's Party (CHP) 2023 congress, removed party leader Özgür Özel, and reinstated his predecessor Kemal Kılıçdaroğlu as interim chair (Al Jazeera).
The sequence matters. The accounts were not blocked for incitement to violence or any concrete, time-bound threat. They were blocked for political speech reacting to a contested judicial intervention in a party's internal democracy — speech that is precisely what free-expression protections exist to shelter.
The strongest case for the state's position
It is worth stating the government's case at its strongest. Every democracy restricts speech that crosses into incitement, defamation, or coordinated manipulation. Turkey, which has lived through a 2016 coup attempt and recurrent terrorism, argues it has a heightened interest in preventing "organized digital activity" from tipping political tension into street violence. The European Convention on Human Rights itself permits proportionate, law-based restrictions on expression to protect public order. If 93 accounts were genuinely orchestrating unrest, a narrowly tailored, court-supervised response would be defensible.
The problem is that almost none of those safeguards are visible here — and the structural record shows why that is not an accident.
A blocking machine that has outgrown the courts
Turkey's Internet Law No. 5651, enacted in 2007, was sold as a tool against clearly defined harms — child sexual abuse material, gambling, narcotics. Successive amendments turned it into a general-purpose censorship engine. The scale is now staggering. According to the Freedom of Expression Association's (İFÖD) EngelliWeb project, Turkey blocked access to more than 311,000 web addresses in 2024 — the highest annual figure on record and up from 240,857 in 2023. The same dataset counts roughly 17,000 X accounts, 75,000 individual posts, and 25,500 YouTube videos restricted in a single year (bianet).
The decisive number is not the total — it is who issues the orders. Of the 311,091 blocks recorded in 2024, only 938 were court-ordered; the telecoms regulator BTK issued 254,130 of them, or about 82%, through administrative decisions. That is the inversion at the heart of Turkey's system: blocking is the default, and a judge's signature is the rare exception. The May 26 action fits the pattern exactly — an executive directorate and prosecutors restricting accounts after "technical and legal reviews," with no published court order, no list of the specific posts, and no named appeal route for the 93 affected users (Stockholm Center for Freedom).
Freedom House rates Turkey "Not Free" on internet freedom, scoring it 31 out of 100 in its Freedom on the Net 2024 report, and documents the steady migration of blocking from courts to administrative bodies, including the removal of 17 VPN services in December 2023 without any court order (Freedom House).
Why proportionality is the whole ballgame
From a pro-innovation, pro-speech standpoint, the objection is not that states may never restrict content. It is that restriction without individualized notice, a specified legal basis, and meaningful judicial review is incompatible with both open expression and a functioning digital economy. When a platform cannot predict whether hosting a citizen's political commentary will trigger an opaque administrative block, it faces an impossible compliance calculus and a strong incentive to over-remove. That chills lawful speech, raises the cost of operating in the market, and pushes users toward circumvention tools the state then races to block — the exact VPN dynamic Freedom House describes.
The timing compounds the harm. Blocking accounts that are reacting to a judicial decision about the opposition's own leadership does not protect public order in any neutral sense; it removes scrutiny from a politically charged ruling at the moment scrutiny matters most. Human Rights Watch called the underlying court decision "a highly unusual interference in a political party's internal party election" that undermines "civil and political rights and Türkiye's democratic process" (Human Rights Watch). Layering account blocks on top of that converts a contested legal outcome into an information blackout.
A pattern, not an episode
This is the same regulatory architecture now expanding into new domains. In April 2026, parliament passed a bill restricting social media for under-15s and requiring identity verification through the state e-Devlet portal — a measure justified by genuine child-safety concerns after a school shooting, but one that, layered onto Law 5651's machinery, hands the state a real-name map of who says what online. Civil-liberties groups have warned globally that age-gating and mandatory verification become surveillance tools when grafted onto weak rule-of-law foundations (EFF).
The lesson for any jurisdiction is narrow and concrete. The legitimacy of online content rules lives or dies on procedure: a specific legal basis, individualized notice, narrow tailoring, and prompt independent review. Turkey built the opposite — a system where 82% of blocks never see a judge — and the 93 accounts are simply its latest output. Proportionate regulation is not the enemy of safety. Censorship by administrative reflex is the enemy of both.