OpenAI Says Iranian Operatives Used AI to Plant Fake Stories in US Media

What happened

OpenAI published a blog post on Thursday disclosing that operatives in Iran used ChatGPT and other models alongside fabricated journalist identities to get roughly 100 articles published across about a dozen outlets worldwide. The articles criticized the US war against Iran. The operatives prompted ChatGPT in Farsi to tweak articles and editorial pitches and to generate social-media comments. After publication, they had ChatGPT mass-produce comments posing as readers reacting to the pieces.

OpenAI said it has banned the ChatGPT accounts involved. Per reporting, the generated comments were specified to portray the US as the aggressor, Israel as an unreliable ally, and Iran as a defiant but resilient victim. None of the outlets that ran the articles appear to be major mainstream news organizations, but some have sizable online followings — London-based Middle East Monitor reportedly has nearly two million Facebook followers. Its director said he would look into the articles OpenAI flagged and take appropriate action.

Key facts

1. Scale of the operation: : Roughly 100 articles across about a dozen outlets, written under seven fabricated journalist bylines; the operatives were based in Iran.
2. How AI was used: : Per OpenAI's disclosure, the operatives prompted ChatGPT in Farsi to tweak articles and editorial pitches and to generate social-media comments masquerading as public reactions.
3. Attribution unclear: : Who ran the operation and whether it is tied to the Iranian government remains unconfirmed — OpenAI said the activity appeared consistent with a commercial actor running a for-hire influence campaign.
4. Wartime backdrop: : Per reporting, this spring a false AI-assisted US intelligence report about a purported Chinese ship carrying nuclear components nearly led the US military to board the vessel.