OpenAI says operatives in Iran used its AI models and fake personas to help place articles critical of the US war against Iran in online outlets, though it cannot say who was behind the effort.
The company said it identified almost 100 articles published or syndicated under the operation's bylines across roughly a dozen online outlets around the world, according to its published report.
OpenAI said on Thursday 8 October 2026 that the operatives used seven fake journalist personas to pitch long-form articles to outlets. The Washington Examiner and Iran International also reported the figure. Crypto Briefing reported that OpenAI called the campaign "Bogus Bylines".
CNN reported that the outlets that carried the material included Middle East Monitor, which is based in London, and LA Progressive. NBC News, as relayed by KPBS, also named Daily Kos.
NBC News reported that Middle East Monitor published at least 21 articles by four fake authors. The reports said the fabrication concerned the identities of the authors and the AI-assisted production of the pieces. The brief available to this newsroom contained no claim that the factual content of the articles was false.
CNN reported that none of the publications appears to be a major mainstream outlet, although some have large online followings. It said Middle East Monitor has almost two million Facebook followers.
According to OpenAI, the operatives also went on to ask ChatGPT to generate comments that could be posted as if they came from people reacting to the articles. The users requested comments on topics such as missile strikes, ceasefires and diplomatic manoeuvring, the company said in its blog post, as carried by CNN.
OpenAI said the generated comments portrayed the US as the aggressor and Israel as an unreliable ally of the US. They described Iran as a victim that was defiant but resilient, the company said. The articles themselves were critical of the US war against Iran, according to CNN.
The attribution to Iran rests on OpenAI's own findings. No independent source has confirmed it, and the company itself was cautious about who was responsible. In its blog post, as quoted by CNN, OpenAI said: "The overall activity appeared consistent with a commercial actor running a for-hire influence campaign, but we are unable to identify the particular actor involved."
CNN also reported that OpenAI could not say whether the operation had links to the Iranian government. No response from Iranian officials had been reported at the time of writing.
Daud Abdullah, director of Middle East Monitor, told CNN that he would look into the articles OpenAI had flagged and take appropriate action. LA Progressive had not responded when CNN asked for comment, according to CNN. The outlets' own checks of the flagged articles had not been reported, and Tehran Telegraph has not independently viewed the articles.
Iran International, an opposition-linked outlet, reported, citing OpenAI's report, that the campaign also repeatedly generated hostile social media responses directed at Iran International itself. OpenAI has not been reported to have responded to that outlet's characterisation beyond the report.
The same OpenAI report covered a separate operation. The company said it also disrupted a Russian operation that used covert think tanks, according to CyberScoop and KPBS.
CyberScoop reported that the stories often drew few comments or social media reactions. Jessica Brandt, quoted by CNN, described it as remarkable that the Iranians had managed to land their content in real media outlets.
OpenAI regularly publishes reports on influence operations that it has disrupted. This one is titled "Disrupting AI-enabled 'false front' operations".
There is a previous case on record. In August 2024, Microsoft detected fake news sites linked to Iran that were aimed at meddling in the US presidential election, according to an NPR report at the time.
OpenAI's latest report does not say whether the outlets have removed the articles, and the company has not named the commercial actor it believes may have been involved. Further details are expected to depend on the outlets' own reviews of the material.
Photo: Jernej Furman from Slovenia / Wikimedia Commons (CC BY 2.0)
