OpenAI Identifies Iranian Group Using ChatGPT to Influence U.S. Election

OpenAI has banned several ChatGPT accounts linked to an Iranian group that attempted to sow division among U.S. voters through AI-generated content. The discovery was part of an internal investigation, which revealed that the group used ChatGPT to create online articles and social media comments centered on divisive topics, including the U.S. presidential election, Israel's war in Gaza, and Israel's participation in the Olympic Games.

AI-Generated Content and its Limited Impact

According to OpenAI, the AI-generated materials were aimed at both progressive and conservative audiences. However, the operation appears to have had minimal impact, with most of the content receiving little to no engagement on social media platforms.

"This operation does not appear to have achieved meaningful audience engagement. The majority of social media posts that we identified received few or no likes, shares, or comments," OpenAI stated in a press release.

The investigation revealed that the group behind the operation published long-form AI-generated articles on websites that falsely posed as news outlets. Additionally, the now-banned ChatGPT accounts were used to craft social media comments, which were often rewrites of existing posts, in both English and Spanish.

"They interspersed their political content with comments about fashion and beauty, possibly to appear more authentic or in an attempt to build a following," OpenAI noted.

Broader Context: Iranian Influence Operations

This revelation from OpenAI comes just a week after Microsoft’s Threat Analysis Center released a report detailing how Iranian-linked groups were using various online tactics to interfere in the U.S. presidential election. The report highlighted the same group, identified as Storm-2035, which was behind the ChatGPT operation.

Microsoft researchers found that Storm-2035 had set up four websites pretending to be American news outlets, amplifying polarizing messages on contentious issues like LGBTQ rights and the Israel-Hamas conflict. The ultimate goal of these efforts, according to Clint Watts, general manager at Microsoft's Threat Analysis Center, was to incite chaos and disrupt the U.S. electoral process.

"Iran is focused as much on just breaking the ability of an election to occur," Watts explained in an interview with NPR.

Limited Reach and Effectiveness

Both Microsoft and OpenAI reported that the websites and AI-generated content linked to Storm-2035 did not gain significant traction online, with low page views and minimal engagement across social media platforms.