OpenAI said on Friday it had removed the accounts of an Iranian group that it said was using its chatbot, ChatGPT, to generate content aimed at influencing the U.S. presidential election and other issues.
According to Open AI, the attack, dubbed “Storm-2035,” used ChatGPT to generate content focused on topics such as commentary on both sides of the US presidential election, the Gaza conflict, and Israel’s participation in the Olympics, which was shared across social media accounts and websites.
An investigation by a Microsoft-backed AI company found that ChatGPT was being used to generate long-form articles and short social media comments.
OpenAI said the tactic appeared to fail to garner meaningful audience engagement.
The majority of the social media posts identified had few or no likes, shares or comments, and there was no indication that the web articles were being shared on social media.
These accounts have been banned from using OpenAI’s services, and the company says it will continue to monitor them for any attempts to violate its policies.
A Microsoft threat intelligence report released in early August said that an Iranian network called Storm 2035, made up of four websites posing as news organizations, is actively engaging with groups of U.S. voters on both ends of the political spectrum.
The effort was built around “polarizing messaging on issues such as U.S. presidential candidates, LGBTQ rights and the conflict between Israel and Hamas,” according to the report.
Ahead of the November 5th presidential election, Democratic candidate Kamala Harris and Republican rival Donald Trump are locked in a fierce battle.
In May, the AI company said it had disrupted five covert influence operations that sought to use its models to carry out “deceptive activities” online.
