OpenAI says fake journalists planted Iran war op-eds

Seven journalists filed roughly 100 articles with about a dozen outlets worldwide. They pitched well, met deadlines, and never asked for a pay rise because none of them was real.

OpenAI has revealed that operatives in Iran used its models and a stable of fake personas to place articles critical of the US war against Iran in several US media outlets, CNN reports. In a blog post, the company said it banned ChatGPT accounts that prompted the tool in Farsi to tweak articles and editorial pitches and generate social media comments.

The pieces weren’t shy about their headlines. One asked, “Is America’s Iran Policy Being Privatised? The Jared Kushner Question”. Another announced “The Iran War and the Collapse of Western Moral Credibility”.

READ MORE: AI never forgets: your past PR crisis is now permanent training data that's hurting your brand reputation

Who ran the articles?

None of the publications that ran the pieces appears to be major mainstream outlets, but some have big enough online audiences to make the effort worthwhile. London-based Middle East Monitor has almost two million Facebook followers.

The Kushner piece ran in LA Progressive, which describes itself as an “independent, justice-driven publication” based in Los Angeles. CNN said it had asked LA Progressive for comment.

Daud Abdullah, director of Middle East Monitor, told CNN he would look into the articles OpenAI flagged and “take the appropriate action”. Another representative for the outlet later said the outlet had taken down the articles and was reviewing its submissions process.

The comments were fake, too

The work didn’t stop at publication. Once the articles were live, OpenAI said the operatives went back to ChatGPT for comments they could post on social media while posing as people reacting to the pieces.

“The users asked for comments on topics such as missile strikes, ceasefires, and diplomatic maneuvering, specifying outputs that portrayed the US as the aggressor, Israel as an unreliable ally of the US, and Iran as a defiant but resilient victim,” the blog post said.

So the operatives wrote the articles, placed them, and also supplied the outraged readers.

Who is behind it?

OpenAI doesn’t know. It also couldn’t say whether the operation is connected to the Iranian government.

“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,” the company wrote.

The campaign is the latest sign of how much generative AI now drives the propaganda battles running alongside the US and Israel’s war with Iran.