An influence operation run from Iran successfully planted roughly 100 articles in media outlets around the world using fake journalist identities and artificial intelligence assistance, OpenAI revealed Thursday. According to the company’s findings, the operators used ChatGPT to edit articles criticizing the American war against Iran, prepare pitches to editors, and generate comments designed to look like they were written by real American users. Following the exposure, the company blocked the accounts used in the operation.
3+According to OpenAI’s report, about a dozen media outlets worldwide published roughly 100 articles submitted under the names of seven fictitious journalists. The network’s operators queried ChatGPT in Persian, asking it to improve article phrasing, tailor publication pitches, and generate additional content designed to promote messages critical of U.S. policy in the war.
Among the article headlines the company identified were: “Is U.S. Policy Toward Iran Going Private? The Jared Kushner Question” and “The War on Iran and the Collapse of Western Moral Credibility.” According to CNN, none of the outlets that published the articles are among the major mainstream news organizations in the United States, but some have significant social media followings.
3+For example, Middle East Monitor, which operates from London, has nearly two million Facebook followers. An article about Jared Kushner and U.S. foreign policy was published on LA Progressive, which presents itself as an independent media outlet working for social justice and based in Los Angeles. CNN reached out to the sites for comment, and Middle East Monitor director Daud Abdullah told CNN he would review the articles flagged by OpenAI and take appropriate action.
According to OpenAI, the activity didn’t end with publishing the articles. After the stories appeared on news sites, the network’s operators asked ChatGPT to generate comments that could be posted on social media as if written by people who had been exposed to content about missile strikes, ceasefires, and diplomatic moves.
3+The company’s report stated that users requested comments portraying the United States as the aggressor, Israel as an unreliable ally of Washington, and Iran as a victim demonstrating determination and resilience. By combining articles on news sites with comments on social media, the operators could attempt to create the impression of authentic public discussion around the messages they were spreading.
Although the activity was identified as operating from Iran, OpenAI emphasized it could not definitively identify the entity behind it or determine whether it was connected to the Iranian government. “The overall activity appeared consistent with a commercial actor running a paid influence campaign, but we are unable to identify the specific actor involved,” the report stated, meaning the findings don’t prove with certainty that this was an operation commissioned or managed directly by the regime.
3+Jessica Brandt, who headed the U.S. Foreign Malign Influence Center from 2023 to 2025 under the Office of the Director of National Intelligence before it was shut down, responded to the exposure in an interview with CNN: “It’s remarkable that the Iranians were able to get their content into real media outlets, thereby expanding the potential reach of their campaign,” adding that “artificial intelligence allows foreign adversaries to better hide their tracks, build more sophisticated fake fronts than they could otherwise create, and launder their content through authentic channels.”