OpenAI Shuts Down Russian and Iranian Influence Operations Using ChatGPT for Fake Journalists and Think Tanks

OpenAI said on Thursday it had dismantled two covert influence operations — one Russian and one Iranian — that used ChatGPT to power propaganda campaigns targeting Ukraine, Latin America, and the war between the United States and Iran. The company banned the offending account clusters in its latest disclosure on the state-sponsored misuse of its AI tools.

The Russian operation, which OpenAI named "Dark Clark," built a fake research platform called the Social Research Center, or SRC, fronted by a fabricated persona named "Mia Clark." The campaign produced fake documents, audio scripts, and misleading material aimed at undermining Ukraine's reputation across Latin America, with some efforts directed at domestic politics in Argentina and Bolivia.

According to OpenAI, the SRC's staff on the ground appeared to have no idea they were working for a Russian group. "The available evidence indicates that the SRC's employees in Latin America were not aware that they were working for a Russian group," the company said. Most of the content was produced as original work by the co-opted staff — a level of sophistication OpenAI said it had not encountered before. "This stands out as the most complex attempt to run a front identity that we've disrupted over the past two and a half years."

OpenAI rated Dark Clark a Category 5 on its six-point influence-operation scale — the highest score it has given to any campaign it has disrupted.

The Iranian operation, dubbed "Bogus Bylines," was quieter but persistent. Seven fabricated journalist profiles, each posing as a Westerner — including from the United States — planted long-form articles in media outlets around the world. OpenAI said it identified almost 100 articles published or syndicated under the operation's bylines across roughly a dozen online outlets, mostly small-to-medium publications focused on international affairs, geopolitics, and Middle East events.

The earliest article appeared in July 2025; the most recent in October 2026. OpenAI said the pace of output increased after the US-Iran war broke out earlier this year. The operation also generated social media comments about the conflict.

One of the personas, "Ervin B. Hoskins," billed himself as an "American freelance writer" and maintained accounts on X and Instagram. Screenshots shared by OpenAI showed X's transparency data linked the account to a "West Asia android app," while Instagram's transparency information placed it in Iran. Both accounts have since been suspended.

The disclosure is the latest illustration of how easily widely available AI tools can be weaponized by state actors to produce propaganda at scale against their adversaries. OpenAI said the banned accounts supported "false front" entities used for geopolitical and conflict-related messaging.

The bust adds to a growing catalog of AI-generated disinformation campaigns uncovered this year. For OpenAI, each disclosure doubles as a warning: the same models powering productivity gains for businesses and developers are equally capable of running convincing influence campaigns — and keeping them out requires constant monitoring.