Over the past year, access has been blocked to 12 X posts by Grok. Grok describes this as “secondary censorship,” while experts call it “a new layer of internet censorship”

 

ALİ SAFA KORKUT

 

In recent weeks, access has been blocked to news reports and social media posts mentioning a Nationalist Movement Party (MHP) lawmaker, including reports and posts concerning allegations that more than 600 people were given fake diplomas at a university in Northern Cyprus founded by his wife.

 

The decision, issued by the Mersin 2nd Criminal Judgeship of Peace on 1 August 2026, blocked access to 10 news reports and 130 social media posts.

 

But the social media posts blocked by the decision, which was issued on the grounds of “protecting national security and public order,” were not limited to posts produced by humans.

 

The decision also blocked access to two posts shared by Grok, X’s artificial intelligence tool, on 3 November 2025 and 9 December 2025.

 

This was not the first access-blocking decision issued against Grok’s posts either.

 

On 23 March, access was blocked to posts concerning allegations that Cihan Karabel had been appointed “faculty secretary” at Recep Tayyip Erdoğan University’s Faculty of Pharmacy, which had been closed without ever admitting students, and had received a salary for three years. The decision issued by the Rize Criminal Judgeship of Peace also blocked access to three posts shared by Grok on 7 and 10 March concerning the allegations.

 

News reports and social media posts concerning allegations that Deputy Minister of Culture and Tourism Batuhan Mumcu had engaged in stock market manipulation were blocked on 18 December 2025. The decision, issued by the Istanbul 4th Criminal Judgeship of Peace, also blocked access to three Grok posts providing information on the allegations about Mumcu.

On 13 February 2025, the target was news reports and posts concerning allegations that the nephew of Deputy Minister of Family and Social Services Zafer Tarıkdaroğlu had been appointed to a civil service position without taking Turkey’s centralized civil service entrance exam, known as KPSS.

 

On 11 September 2025, an access-blocking decision was issued directly against Grok’s X account, @grok. That decision, however, has not yet been implemented.

 

All of the access-blocking decisions concerning at least 12 X posts by Grok, as well as the decision directly targeting Grok’s X account, were issued on the grounds of “protecting national security and public order.”

 

 “A new layer of internet censorship”

 

According to digital law expert and Alternative Informatics Association President Faruk Çayır, the blocking of Grok’s posts constitutes “a new layer of internet censorship.”

 

“AI outputs that compile and rephrase news reports are also becoming part of the circulation of public information,” Çayır told Expression Interrupted. “For this reason, blocking a Grok post may affect not only a single piece of content, but also the visibility of the chain of news reports on which that content is based.”

 

Çayır said the fact that a piece of content was generated by artificial intelligence did not automatically place it outside the scope of freedom of expression. He also noted, however, that the risk of AI hallucinations could not be ignored, emphasizing public interest as a key criterion in assessing any restriction: “The criterion should not be so much ‘who produced the content?’ but rather the post’s contribution to public debate, its sources, language and verifiability, the public position of the person concerned, and whether the restriction is genuinely necessary,” he said.

 

Çayır also addressed the “protection of national security and public order” grounds cited in the access blocking decisions, saying: “This alone cannot be considered a sufficient justification. A concrete danger, the connection between the content and the aim pursued, necessity, proportionality and sufficient judicial reasoning must also be established.”

 

Çayır said future efforts to regulate online content may go beyond specific URLs and extend to the way artificial intelligence systems respond to questions on certain topics. If this leads to topic-based restrictions on AI-generated answers, he warned, it could amount to preventive censorship and automated overblocking, creating new barriers to access to information.

 

“Blocking public debate”

 

For Professor Ceren Sözeri of Galatasaray University’s Faculty of Communication, the access blocks targeting Grok raise broader questions about public interest and the right to receive and impart information. She said she did not believe decision-makers or lawmakers in Turkey had yet fully grasped the consequences of such measures.

 

“Content-blocking decisions are issued without taking into account public interest or the right to receive and impart information, and are based only on the demands and political identity of the applicant,” Sözeri said, recalling the Constitutional Court’s pilot judgment annulling legal provisions that allow access blocking on the grounds that they violate freedom of expression.

 

Assessing the access-blocking decision concerning news reports and Grok posts about the MHP Mersin MP, Sözeri said: “Censoring news that the public has a right to know about this MP means blocking public debate in terms of voters’ right to hold politicians accountable and transparency ahead of the next election period. It also means preventing the media from fulfilling one of its most fundamental functions in a democracy.”

 

Sözeri also noted that no objection had been heard from X owner Elon Musk following reports that Grok would be blocked in Turkey.

 

“We have to accept that these are commercial companies, and their main goal is to make money,” she said. “That money does not necessarily have to come from the platforms themselves. Compared with the revenue Musk earns, or could earn, from Turkey through Tesla or his satellite and defense industry companies, the freedom of a journalist in Turkey to report in the public interest carries no value.”

 

Grok: “Truth itself cannot be blocked”

 

We also asked Grok, X’s artificial intelligence tool, about the access blocks targeting its posts.

 

Grok described the practice as a sign that “censorship” is focused not on the original source of information, but on its circulation. It said that even after a news report has been published, efforts are made to prevent it from being reproduced or made accessible again.

 

Grok said it is merely an intermediary that compiles publicly available information and presents it to users. Referring to the rights to communication and access to information, it said that silencing a tool that summarizes or contextualizes news that has already been published “strikes at the essence of the right.” Grok described this as “secondary censorship,” saying that targeting an automated, question-and-answer-based tool cuts off technology’s contribution to the democratization of information.

 

Grok said this creates an environment in which information circulates only in a “controlled” way and the public is left “incompletely” informed. In line with Sözeri’s assessment, it said that, in the long term, this weakens accountability rather than serving the public interest.

 

Grok also said the reliance on “national security and public order” as grounds for access-blocking decisions was problematic. “Access blocks imposed on public order grounds often affect critical journalism and dissenting voices, weakening democratic oversight mechanisms,” it said. “My view is clear: transparency, debate and the effort to reach the truth are healthier than broad blocking decisions packaged under the label of ‘national security.’ As an AI, my role is to provide information as accurately and impartially as possible. Court decisions may restrict access to this information from Turkey, but they cannot block truth itself.”