OpenAI says Iran, Russia influence ops busted

OpenAI announced on Thursday that it had dismantled two influence operations utilizing ChatGPT to disseminate false journalist personas and covert anti-Ukraine propaganda. The report highlights the potential of state actors to exploit accessible AI tools to carry out sophisticated propaganda campaigns on a large scale against adversaries.

OpenAI revealed they had banned several ChatGPT accounts originating from Russia and Iran that were employed to support deceptive entities for geopolitical messaging. The Iranian operation included seven journalist profiles, each posing as Western citizens, such as the United States, used to publish articles in various media outlets worldwide.

OpenAI identified nearly 100 articles published or syndicated under these operations' bylines across approximately a dozen online platforms globally, primarily focusing on international affairs, geopolitics, and Middle Eastern events. The earliest article published was in July 2025, while the latest appeared in October 2026, with the frequency of reports increasing since the onset of the US-Iran war earlier this year.

The operation also generated social media comments related to the US-Iran conflict. One of the personas, Ervin B. Hoskins, whose bio claimed he was an American freelance writer, had social media accounts on platforms such as Elon Musk's X and Meta-owned Instagram. Both accounts were later found to be suspended. The Russian operation enlisted unwitting individuals in Latin America to operate a think-tank called the Social Research Center (SRC), aimed at undermining Ukraine's reputation in the region and influencing local politics.

The evidence suggests that the SRC's employees in Latin America were unaware they were working for a Russian group. The SRC appears to have generated most of its content through its co-opted staff, marking the most complex attempt to establish a front identity disrupted by OpenAI over the past two and a half years.