Front groups and fake journalists: OpenAI bans networks from Russia and Iran
A Russian operation targeted Latin America with a fake research institute, an Iranian one slipped about 100 articles into real media. OpenAI rates the Russian case higher than any before.
Symbolic image · Photo: Jernej Furman from Slovenia / Wikimedia Commons, CC BY 2.0 (cropped and resized)
OpenAI has banned two covert influence operations from Russia and Iran that used ChatGPT alongside traditional methods to run so-called false fronts, according to a report published on 8 October. They are two of the 30 covert operations the company says it has exposed in roughly two and a half years. AI can give such operations “greater scale, efficiency, linguistic fluency, and editorial ability,” OpenAI writes.
The Russian operation, dubbed “Dark Clark,” targeted Latin America: through a supposed research institute and an invented persona named “Mia Clark,” it spread fake official letters to schools in Peru and Ecuador as well as fake audio recordings, for example to damage Ukraine. According to OpenAI, Latin American staff were apparently unaware of the Russian control. Open-source researchers link some of the fakes to “Politology,” which is regarded as a successor to the Wagner Group. On the Breakout Scale that OpenAI uses to rate the reach of such campaigns, it is the first operation OpenAI has disrupted to reach Category 5 on the six-point scale.
The Iranian operation, “Bogus Bylines,” used seven invented Western journalists. Between July 2025 and October 2026, almost 100 of their articles on the US-Iran conflict appeared in about a dozen outlets. OpenAI believes it was likely a commercial for-hire service but could not identify the client. The accounts used VPN connections.