OpenAI Bans Russia- and Iran-Linked Influence Operations That Used ChatGPT to Run Fake Journalists and a Fake Think Tank
OpenAI disclosed that it banned two covert, AI-enabled influence operations — one linked to Russia, one to Iran — that used ChatGPT to help run fabricated news and research personas.
Details
- “Dark Clark” (Russia-linked): a cluster of accounts operating in Russian via VPNs, with evidence tying operators to “Politology”/“La Compania,” described as a reported successor to the Wagner Group and entities founded by Yevgeniy Prigozhin; the operation created the “Social Research Center” (SRC), a fake Latin American think tank with unwitting local employees unaware they served Russian interests
- “Bogus Bylines” (Iran-linked): operators prompting in Persian while generating English and Persian content, resembling a commercial for-hire influence campaign; deployed seven fabricated journalist personas backstopped by social media accounts
- How ChatGPT was used: Dark Clark used it mainly to write internal activity reports and refine language for regional authenticity; Bogus Bylines used it to refine long-form articles and editorial pitches, generate social media comments, and refine internal reports
- Targeted narratives: Dark Clark targeted Ukraine’s reputation in Latin America and domestic politics in Argentina, Bolivia, and Ecuador, including fabricated stories about Argentina’s president and anti-Polish sentiment via a historical controversy; Bogus Bylines pushed anti-war narratives framing the US as the aggressor in US-Iran tensions and criticized Iran International
- Scale: using the Brookings Institution’s Influence Operations Breakout Scale (1-6), Dark Clark reached Category 5 (nearly 100 articles across roughly a dozen outlets, provoking official denials and comment from Polish MEPs) and Bogus Bylines reached Category 4 (almost 100 articles across small-to-medium outlets)
- Detection methods: account metadata analysis, prompt clustering, open-source research, and cross-referencing with fact-checks and official denials
What happened next
OpenAI noted both operations resembled “pre-AI age” influence tactics but used AI for scale, efficiency, and linguistic fluency, and argued that publicly disclosing false-front operations tends to neutralize them — citing how earlier campaigns like “Alice Donovan” and “PeaceData” ceased operating once exposed; the disclosure extends OpenAI’s ongoing pattern of periodic threat-intelligence reports following its earlier model-distillation campaign takedown.