OpenAI Exposes Iranian Influence Operation Using AI to Plant Fake Articles
OpenAI uncovered an Iranian influence operation that used ChatGPT and fake journalist personas to plant 100 articles across global media outlets to sway public opinion.
OpenAI has dismantled an influence operation originating from Iran that successfully planted approximately 100 articles in global media outlets using fabricated journalist identities and artificial intelligence tools.
Iranian Influence Campaign via ChatGPT
According to the report published by OpenAI, the operators leveraged ChatGPT to edit articles criticizing US policy, draft publication pitches, and generate social media comments designed to mimic authentic American users. Following the detection, OpenAI blocked the accounts associated with the malicious activity.
Approximately a dozen media outlets worldwide published the content submitted under the names of seven fictitious journalists. The network operators utilized the Persian-language interface of ChatGPT to refine text, adapt pitches, and promote critical messaging regarding Western policies.
Media Placement and Fake Personas
Among the article headlines identified by OpenAI were pieces questioning US foreign policy and discussing the moral credibility of the West. While major mainstream American news networks were not targeted, secondary outlets with significant social media followings, such as the London-based Middle East Monitor and the Los Angeles-based LA Progressive, published the material.
"It is remarkable that the Iranians managed to get their content into real media outlets, thereby expanding the potential reach of their campaign," noted Jessica Brandt, former head of the US Center for Threat-Informed Foreign Influence.
OpenAI stated that the operation also generated coordinated inauthentic comments across social media platforms following the publication of the articles, framing the United States as the aggressor and portraying Iran as resilient.