The Iranian deception operation: 20 US media outlets fell for it

An Iranian influence campaign exploited ChatGPT to spread about 100 fake articles and opinion pieces in at least 20 US media outlets. The operators used fictitious identities, such as "Erwin Hoskins," whose name also appeared on major sites. OpenAI uncovered the activity and blocked the accounts.

An Iranian influence campaign exploited ChatGPT to spread fake content in at least 20 US media outlets. As part of the operation, about 100 articles and opinion pieces were published under fictitious identities, such as "Erwin Hoskins," who appeared in the "Port St. Joe Star" and on major sites like "Daily Kos" and "Middle East Monitor." OpenAI revealed that Hoskins' profile picture was forged using the company's AI tool. The activity was described as a commercial-for-hire operation, focusing on spreading critical content against US military activity in Iran, while highlighting negative impacts on American civilians. The operators from Iran sent instructions to ChatGPT in Persian, asking it to write articles in English and tailor them to the requirements of specific media outlets. OpenAI blocked the involved accounts.