OpenAI Discloses Two AI-Assisted Covert Influence Operations: Using AI to Create "Fake Identities" and Spread Content

Odaily reports that OpenAI has released a report stating it recently banned two groups that used its models to conduct covert influence operations (IO). These operations combined AI with traditional techniques to create "fake front" entities and fake identities. OpenAI stated that the Russia-linked operation "Dark Clark" controlled a fake think tank called the "Social Research Center (SRC)," and is suspected of using ChatGPT to write internal reports, generate content, and conduct influence activities in Latin America. OpenAI assessed this operation as Level 5 on the IO Breakout Scale, marking the first time it has identified a case reaching this level since it began disclosing such operations.

OpenAI stated that AI has not changed the fundamental patterns of these influence operations, but has enhanced their scale, efficiency, language capabilities, and content production capacity, making it easier for attackers to create fake identities, fabricate content, and disseminate information to real media channels.