OpenAI caught Russians and Iranians using ChatGPT for influence campaigns

OpenAI disclosed on Thursday that it had dismantled two influence operations, one tied to Russia and the other to Iran, which utilized ChatGPT to disseminate deceptive journalist personas and promote anti-Ukraine propaganda. The company emphasized that its report underscores the ease with which state actors can leverage readily available AI tools to disseminate intricate propaganda aimed at adversaries on a large scale.

OpenAI claimed it had shut down a network of ChatGPT accounts originating from Russia and Iran, which were employed to support false front entities for geopolitical and conflict-related messaging. The Iranian campaign involved seven journalist profiles, each pretending to be from Western nations including the United States, designed to publish long-form articles in media outlets globally.

OpenAI stated that almost 100 articles were published or syndicated under these profiles across roughly a dozen online platforms worldwide, primarily focused on international affairs, geopolitics, and Middle Eastern events. The earliest article detected by OpenAI was published in July 2025, while the latest appeared in October 2026, with publishing frequency rising following the commencement of the US-Iran conflict earlier in the year. The operation also generated social media comments on matters related to the US-Iran war.

One of the personas, Ervin B. Hoskins, who claimed to be an American freelance writer, maintained social media accounts on various platforms such as Elon Musk's X and Meta-owned Instagram. X's transparency data revealed the account was linked through a West Asia android app, while Instagram's transparency data indicated the account was based in Iran.

Both accounts were reportedly suspended. The Russian operation exploited unsuspecting individuals in Latin America to run a think-tank dubbed the Social Research Center (SRC), which was utilized for covert activities aimed at tarnishing Ukraine's image in the region and influencing local politics. OpenAI noted that the SRC's Latin American employees were unaware they were working for a Russian group.

The SRC largely produced original content via its manipulated staff, marking one of the most intricate attempts to establish a front identity disrupted by OpenAI over the past two and a half years.