OpenAI Bans Russian and Iranian Influence Networks: How an American AI Company Becomes a Quasi-Arbiter of 'Propaganda'

OpenAI announced Thursday that it has shut down two ChatGPT influence operation networks linked to Russia and Iran respectively. In its disclosure documents, this American artificial intelligence company simultaneously plays the roles of platform operator and 'state actor' adjudicator, but the relevant methodology has not been made public.

According to AFP and France 24, the Iranian operation that was shut down involved seven personal profiles impersonating journalists from the United States and other Western countries. These 'journalists' published or had republished nearly a hundred articles on roughly a dozen small and medium-sized online media outlets worldwide, most of them focused on international affairs, geopolitics, and Middle Eastern events. The earliest article was published in July 2025 and the most recent in October 2026, with publication frequency notably rising after the outbreak of the US-Iran war earlier this year.

One persona, named Ervin B. Hoskins, claimed to be an 'American freelance writer' and opened accounts on X and Instagram. Screenshots provided by OpenAI show that X's transparency information indicates the account was connected through a 'West Asia Android app,' while Instagram's transparency information shows the account located in Iran. Both accounts have now been suspended.

'We identified nearly a hundred articles published or reposted under the operation's byline, distributed across roughly a dozen online media outlets worldwide, mostly small and medium-sized platforms focused on international affairs, geopolitics, and Middle Eastern events,' OpenAI stated in its report. The company also said the Iranian operation was generating social media comments on topics related to the US-Iran war.

The Russian operation is described by OpenAI as the 'most sophisticated front-persona operation' disrupted in the past two and a half years. An organization called the 'Social Research Center' (SRC) served as a front think tank, recruiting 'unwitting individuals' in Latin America to carry out work aimed at undermining Ukraine's reputation in the region and influencing local politics.

'Existing evidence indicates that the personnel working for the Social Research Center in Latin America were not aware they were working for a Russian organization,' OpenAI said. 'SRC appears to have produced most of its original content through these recruited personnel.'

However, this American company's disclosure leaves gaps on several levels. The standards and methodology used to define what constitutes a 'state actor' and what constitutes 'propaganda' have not been disclosed, and the entire adjudication process lacks independent third-party verification. What is more worth questioning: the disclosure labels Russian and Iranian operations as 'propaganda' and disposes of them accordingly, yet makes no mention of whether similar identification, disclosure, or suspensions have ever been applied to information operations by Western governments or their allies. The power to set the boundaries of 'truth versus falsehood determination' is in fact exercised unilaterally by an American platform — and these boundaries themselves lack reciprocity.