OpenAI reports disrupting two influence operations built around false fronts

Operations using fictitious authors

OpenAI says it banned two influence operations that it attributes to actors from Russia and Iran. The Iranian network used seven fabricated journalist personas to pitch articles to small and midsize outlets and also generated batches of comments. The Russia-origin operation allegedly enlisted unwitting people in Latin America to operate a deceptive think tank and produced fake leaked documents and audio scripts.

The company’s models were also used for internal reporting, not only outward-facing content. OpenAI rated the Russia-origin operation’s potential reach at Category 5 of 6 on the Breakout Scale, the first disrupted network it has placed at that level; it rated the Iran-origin activity Category 4.

AI’s role in the reported campaigns

OpenAI describes AI use for producing text, creating fictitious authors and preparing internal reports. People organised the campaigns. Network attribution and reach estimates come from the company’s report.