Iranian Group Used ChatGPT to Create Divisive Articles

An Iranian group used ChatGPT to create content that was intended to influence the US presidential election and spread misinformation online, OpenAI said Friday. The company, which said the offending accounts have been banned, said its investigation showed the group, which it dubbed Storm-2035, used ChatGPT to create longform articles and social media posts. The subject matter was wide-ranging: Topics included the US presidential election, the situation in Gaza, Israel's presence at the Olympics, and Scottish independence, reports the New York Times.

Though the content was then shared on websites and via social media accounts on X and Instagram, OpenAI says it appears the content didn't attract many eyeballs: Most social posts had very little interaction, and the web articles didn't appear to have been shared by readers, Reuters reports. "We did not see signs that it was getting substantial engagement from real people at all," said OpenAI investigator Ben Nimmo.

The Washington Post reports on two of the websites pinpointed by OpenAI as having ChatGPT-generated content: One called Teorator, which purports to be "your ultimate destination for uncovering the truths they don't want you to know," hosted articles critical of Tim Walz. Another, called Even Politics, featured articles critical of Donald Trump. (More ChatGPT stories.)