OpenAI said it has removed ChatGPT accounts originating in Russia and Iran that were used for influence operations in several countries.
The company announced the move on Thursday 8 October 2026, describing two separate operations, one from Russia and one from Iran, according to reports by AFP via France 24 and Bloomberg Law.
All attribution of the operations to Russia and Iran comes from OpenAI. No independent confirmation had been reported, and no response from the Iranian or Russian governments had been reported in the coverage reviewed by Tehran Telegraph.
According to AFP, the Iranian operation used seven purported journalist profiles, each claiming to be from a Western country, including the United States. The profiles were used to place long-form articles in outlets around the world.
OpenAI said the articles appeared in outlets that were "small to medium outlets, generally focused on international affairs, geopolitics, and events in the Middle East". WION reported that the earliest article OpenAI identified dates from July 2025.
AFP reported that one of the personas, named Ervin B. Hoskins, described himself as an American freelance writer. He had accounts on X and Instagram, which appeared to have been suspended.
The Russian operation worked differently, according to AFP and NPR. It used a purported think tank, the Social Research Center (SRC), which OpenAI said sought to undermine Ukraine's reputation in Latin America and to influence local politics.
OpenAI said the available evidence indicates that SRC staff in Latin America did not know they were working for a Russian group, AFP reported. NPR reported that the accounts mostly used ChatGPT to write reports addressed to unspecified superiors.
Bloomberg Law reported that the Russian operation also created fake "leaked" documents and scripts for audio, some of which were spread online. NPR said the findings shed light on campaigns not previously attributed to Russia.
OpenAI rated the Russian operation category 5 and the Iranian operation category 4, according to Bloomberg Law. It said the Russian operation was the first at category 5 since the company began reporting.
CyberScoop reported that the scale runs to 6, and The Hill reported that the Russian operation had the greatest reach OpenAI has recorded. The ratings are OpenAI's own assessment.
NPR reported that OpenAI found campaigns which placed false narratives in established media appeared to reach wider audiences than those relying on fake social media accounts.
NPR also reported that, in internal reports produced with ChatGPT, the operatives said they had got narratives picked up by news outlets. In one case, according to those reports, this prompted a response from a member of the European Parliament. NPR attributed this to the operations' own reports, and it has not been independently verified.
The geographic scope of the campaigns differed. The Russian operation targeted Latin America, while the Iranian operation used personas posing as Western journalists, some purporting to be in the United States, and published in outlets worldwide.
The announcement follows an earlier OpenAI report, published on 25 August 2026, which covered a Russia-origin cluster promoting the "International Burke Institute", a self-described expert community based in Israel, and a "sovereignty" index. That report is available on OpenAI's website.
OpenAI regularly publishes reports on disrupting covert influence operations, according to OpenAI and Bloomberg Law.
Photo: Jernej Furman from Slovenia / Wikimedia Commons (CC BY 2.0)
.jpg){kind=link}