OpenAI says it banned Russian- and Iranian-origin ChatGPT account clusters tied to covert campaigns that used fake personas and media placements.
OpenAI said on October 8, 2026, that it banned two groups of ChatGPT accounts associated with covert influence efforts originating in Russia and Iran. The campaigns used fabricated identities and organizations to place geopolitical messaging with audiences, alongside more conventional online tactics.
The Russian-linked effort, which OpenAI called “Dark Clark,” targeted countries across Latin America. The company says the operators used ChatGPT mostly to prepare internal reports, while also generating some campaign material. Their activity centered partly on a purported research organization, the Social Research Center, which OpenAI says was presented as an independent entity. OpenAI found evidence of more than 60 articles on its website and said local staff appeared to have been unaware of the operation’s Russian connection. It rated the campaign Category 5 on the six-point Breakout Scale, citing evidence that some claims circulated widely enough to prompt fact checks and official responses.
The Iranian-origin operation, called “Bogus Bylines,” used seven invented journalist identities to pitch articles about the US-Iran conflict. OpenAI said it identified nearly 100 articles published or syndicated under those bylines across roughly a dozen outlets. The campaign also used ChatGPT to produce batches of social media comments, though OpenAI found little evidence that those replies drew substantial engagement. It assessed the article-placement activity as Category 4 and the commenting activity as Category 2.
OpenAI cautioned that operators’ internal reports overstated or misattributed some results, so their claimed reach could not be accepted at face value. The company said it shared information about both cases with relevant authorities. It also noted that AI helped with tasks such as drafting reports, refining articles and adapting language, while the broader operations relied on established methods—including deceptive identities and efforts to place material in outside publications.