OpenAI has suspended ChatGPT accounts linked to Russian and Iranian influence operations accused of using artificial intelligence to spread propaganda, manipulate political discussions and promote misleading narratives about international conflicts.
The company disclosed the action on Thursday, October 8, saying the two operations used its technology alongside other methods to disguise their origins and distribute geopolitical messaging through seemingly legitimate organisations and media outlets.
According to OpenAI, the Iranian operation created seven fake journalist identities to pitch lengthy articles to small and medium-sized online publications worldwide. The Russian-linked campaign, meanwhile, allegedly recruited unsuspecting individuals in Latin America to operate a purported think tank.
The findings highlight growing concerns about the use of artificial intelligence tools to support covert influence campaigns and introduce misleading content into established information channels.
OpenAI identified nearly 100 articles published or syndicated under the Iranian operation’s fake bylines across roughly a dozen online publications. The outlets generally covered international affairs, geopolitics and developments in the Middle East.
The earliest article identified during the investigation appeared in July 2025, while the latest was published in October 2026. OpenAI said the volume of articles increased significantly following the outbreak of the US-Iran conflict.
The operation also used ChatGPT to refine articles, prepare pitches to editors and generate batches of social media comments, many of which focused on the conflict and its consequences for Americans.
One of the fabricated journalist identities was Ervin B. Hoskins, who described himself online as an American freelance writer. The identity was associated with accounts on several social media platforms, including X and Instagram.
OpenAI said transparency information on the accounts indicated links to Iran. The company also identified other fabricated bylines used to publish articles, with some of the content appearing in publications that attracted substantial online audiences.
The operation appeared consistent with a commercial influence campaign offering its services for hire, although OpenAI said it could not identify the specific actor responsible. The company added that it had shared information about the case with relevant authorities.
In the separate Russian-linked operation, OpenAI identified a network of accounts used to support activities targeting countries across Latin America.
Much of the campaign appeared aimed at undermining Ukraine’s reputation in the region, while some activities sought to influence domestic political developments, particularly in Argentina and Bolivia.
The operators allegedly used ChatGPT to prepare internal reports, produce campaign material and document activities linked to a purported research organisation known as the Social Research Center (SRC).
OpenAI said the Russian-linked operators appeared to control the organisation through a fabricated identity named Mia Clark. Evidence from internal reports suggested they made decisions concerning staffing, salaries and recruitment.
The company said individuals recruited in Latin America appeared unaware that they were working for a Russian-linked operation. Unlike some other covert campaigns, the SRC reportedly produced much of its original content through these recruited workers.
OpenAI also identified fabricated documents, audio scripts and misleading stories associated with the network. Some claims reportedly circulated online widely enough to prompt fact-checks and official denials.
The company classified the Russian operation as a Category 5 threat on the Influence Operations Breakout Scale, which measures the reach and impact of covert campaigns on a scale of one to six. It rated the Iranian operation at Category 4.
OpenAI said both campaigns demonstrated how artificial intelligence could make covert influence operations more efficient by helping operators produce content, maintain deceptive identities and exploit the trust audiences place in legitimate media outlets and organisations.
Goodness Anunobi
Follow us on:
