An influence campaign originating in Iran planted about 100 articles in media outlets worldwide by creating fictitious journalist personas with the help of artificial intelligence, OpenAI revealed last Thursday. According to thecompany’ss findings, the operators used ChatGPT to draft and edit commentaries criticizing the U.S. military campaign against Iran, prepare publication pitches, and generate comments engineered to appear as though written by American social media users. Following the investigation, OpenAI banned the accounts used in the operation.
According toOpenAI’ss published report, roughly a dozen media outlets worldwide published approximately 100 articles submitted under the names of seven fabricated journalists. The network’s handlers prompted ChatGPT in Persian to polish article drafts, tailor pitch emails to editors, and create supplementary content designed to advance sharp critiques of U.S. wartime policy.
Among the article headlines identified by OpenAI were: “Is US Iran policy going private? The Jared Kushner question”” and”“The war on Iran and the collapse of theWest’ss moral authority”” According to CNN, none of the platforms that ran the submissions rank among mainstream flagship news outlets in the United States. Yet, several maintain substantial followings across social media platforms.
For instance, the London-based outlet Middle East Monitor commands nearly two million followers on Facebook. At the same time, an article discussing Jared Kushner and American foreign policy appeared on LA Progressive, an independent social justice website based in Los Angeles. CNN reached out to the outlets for comment, and Daud Abdullah, director of Middle East Monitor, told CNN he would review the articles flagged by OpenAI and take appropriate action.
According to OpenAI, the covert operation did not stop with publishing the articles. Once the pieces went live on news websites, the network directed ChatGPT to draft synthetic comments designed for distribution across social platforms, mimicking ordinary users reacting to content concerning missile strikes, ceasefires, and diplomatic moves.
OpenAI’s report noted that users requested comments framing the United States as the aggressor, depicting Israel as an untrustworthy ally to Washington, and presenting Iran as a steadfast, resilient victim. By pairing published articles with coordinated social commentary, the handlers tried to create the illusion of authentic grassroots public discussion around their messaging.
Although the activity was traced to Iran, OpenAI noted it could not definitively identify the exact entity behind the network or determine whether it operated under the direct auspices of the Iranian government.” he overall activity appeared consistent with a commercial actor running a paid influence campaign. Still, we are unable to identify the specific actor involved,” the report stated, indicating the findings do not decisively prove whether the regime directly commissioned or managed the operation.
Jessica Brandt, who led the Foreign Malign Influence Center under the U.S. Office of the Director of National Intelligence between 2023 and 2025 before its closure, addressed the findings with CNN: “It is remarkable that the Iranians managed to place their content into actual media outlets, thereby broadening the potential reach of their campaign.” Bran”t added: “AI allows foreign adversaries to conceal their footprints better, build far more sophisticated fronts than they could otherwise muster, and launder their messaging through authentic distribution channels.”