OpenAI says it disrupted Russian and Iranian influence operations that used fake identities to place material in legitimate online publications. The activity differed from a conventional social-media spam campaign because some fabricated stories reached real outlets and political audiences.
The Russian operation, which OpenAI calls Dark Clark, targeted Latin America with content intended to discredit Ukraine and destabilize local politics. Operators used a fictitious persona to control a think tank and may have involved local staff who did not know the wider campaign. False audio and documents prompted fact-checks and official denials in Ecuador and Peru.
OpenAI rated Dark Clark at five on its six-level Breakout Scale, its highest assessment in two and a half years of public reporting, because politicians responded to the material. The rating reflects observed reach and reaction, not proof that the campaign changed political outcomes.
A separate Iranian operation, Bogus Bylines, operated seven fake journalist profiles and placed nearly 100 articles about the US-Iran conflict in publications around the world. Its associated social posts attracted little attention. According to OpenAI, both groups used ChatGPT accounts mainly for internal reporting and adapting propaganda across languages. The company banned the identified accounts; removing those accounts does not erase already published stories or the false identities behind them.