Iranian campaign planted fake articles in real U.S. publications using ChatGPT
“The bill will follow Americans beyond the fighting itself,” an opinion piece in the Port St. Joe Star, a tiny Florida weekly newspaper, warned of the Iran war last month.
TL;DR
- OpenAI identified an influence campaign originating from Iran that used ChatGPT to create fake articles.
- Around 100 AI-generated articles were placed in at least 20 U.S. news outlets under fictional bylines.
- The articles often focused on negative impacts of the Iran war on Americans and were written in English after prompts in Persian.
- Some articles featured AI-generated author photos, and the operation seemed to be commercially run.
- News outlets like the Port St. Joe Star, Middle East Monitor, and others were impacted, with some taking down the content.
- PeaceVoice, a think tank that pitched an article by a fake byline, has since reinforced its no-AI policy.
- OpenAI has banned the accounts involved and is blocking ChatGPT access from Iran, sharing information with authorities.