OpenAI Axes Iranian Group’s ChatGPT Accounts Over Election Interference

On Friday, OpenAI disclosed that it had shut down accounts operated by an Iranian group known as Operation Storm-2035. This group used OpenAI’s ChatGPT chatbot to create various types of content aimed at influencing the U.S. presidential elections and other topics. This content, which included commentary on U.S. election candidates and issues like the conflict in Gaza and Israel’s participation in the Olympic Games, was then posted on social media platforms and websites.

However, OpenAI’s investigation revealed that these efforts did not engage a significant audience, with many posts receiving little to no interaction in terms of likes, shares, or comments. Furthermore, there was no evidence to suggest that the articles generated were widely shared on social media platforms.

Following the findings, OpenAI has banned the implicated accounts from accessing its services and continues to monitor for any new attempts to breach its usage policies.

Earlier in the month, a Microsoft threat-intelligence report identified Storm-2035 as an Iranian network behind four websites posing as news outlets. These sites were actively targeting U.S. voter groups with polarizing content about the presidential candidates, LGBTQ rights, and the Israel-Hamas conflict.

The upcoming presidential election on November 5 features Democratic candidate Kamala Harris and Republican Donald Trump in a close contest.

OpenAI also noted that this incident was one of five similar operations it disrupted in May, all intended to misuse its AI models for deceptive online activities.