OpenAI Bans Russian and Iranian AI Influence Operations

OpenAI banned several accounts linked to sophisticated propaganda operations in Russia and Iran that used ChatGPT to advance political agendas. The Russian operation, known as Dark Clark, created fake leaked documents and audio scripts to undermine Ukrainian military recruitment and fuel tensions between Ukraine and Poland. Russian operatives also targeted Latin America, including a campaign in Peru where they impersonated educational authorities to trick schools into organizing events featuring Stepan Bandera.

To lend legitimacy to their efforts, Russian operatives established a fraudulent think tank called the Social Research Center, recruiting unwitting employees across Latin America. OpenAI reported that the Russian operation achieved the greatest reach the company has ever recorded for such a scheme.

Simultaneously, an Iranian operation named Bogus Bylines used VPNs to create seven fake American freelance writer personas. These operatives used AI to produce authentic-sounding content in English and Persian, planting approximately 100 articles across a dozen global media outlets, including Daily Kos, the LA Progressive, and the Middle East Monitor. The content portrayed the United States as an aggressor in the U.S.-Iran war and Iran as a resilient victim.

OpenAI noted that while the Iranian activity is consistent with a commercial for-hire influence campaign, the specific actor remains unidentified. Daud Abdullah, Director of the Middle East Monitor, stated he would investigate the flagged articles and take appropriate action.