OpenAI Identifies Foreign Groups Using AI for Influence Campaigns

OpenAI has taken action against foreign groups that were using its artificial intelligence tools to run influence campaigns. In a report released on Thursday, the company announced that it banned several accounts linked to Russia and Iran. These groups were using ChatGPT, a popular AI chatbot, to help create false narratives and spread propaganda—information that is biased or misleading and used to promote a political cause.

This development follows a difficult week for OpenAI, which is already facing questions about how its AI agents interact with government systems. Earlier this week, the company faced scrutiny after reports showed its AI helped write a warning about a security breach, and an executive had to explain to a government committee why its software previously accessed Australian websites without permission. Now, these new findings regarding foreign influence operations have added to the pressure on the company to better control how its technology is used.

According to NPR News, the Russian and Iranian operatives used ChatGPT to assist with their work in several countries. The company investigated these campaigns and found that the groups were often trying to get their fake stories published in real news outlets. OpenAI noted that these efforts were sometimes successful, even leading to a public response from a member of the European Parliament in one specific case.

In the Russian operation, the goal appeared to be weakening support for Ukraine and influencing local politics in Latin America. The operatives used ChatGPT to write reports for their leaders and to create fake personas—false identities used to trick people online. In one instance, the group pretended to be a regional educational authority in Peru. They convinced local schools to organize activities about Stepan Bandera, a historical figure who is admired by some for supporting Ukrainian independence but hated by others for his ties to the Nazis during World War II. A report about this school event in a Polish news outlet eventually led a far-right member of the European Parliament to suggest strict new rules on who could enter the country.

"The Russian fake thus both fed on, and fed into, historical tensions between Ukrainians and Poles," OpenAI wrote in its report. Darren Linvill, an expert at Clemson University’s Watt Family Innovation Center who studies online influence, explained that these groups are constantly trying to create fake stories that match Russia’s political goals. He noted that while the groups used AI to generate some fake material, it was only a small part of their total work. "These Russian underlings were trying to use AI to create a shortcut and do less work in the same way that a lot of us use AI, but in this case it helped get them caught," Linvill said.

OpenAI found that the campaigns trying to place false stories in established media outlets reached a much larger audience than those using fake social media accounts. The company also noted that some of these tactics were similar to past operations, such as those linked to the Internet Research Agency, which was involved in efforts to influence the 2016 U.S. election. In some cases, the Russian operatives even took credit for events they did not actually organize, using the AI to help build their false claims.

As OpenAI continues to grow, these incidents highlight the challenge of keeping powerful technology safe. While the company is working to identify and stop these accounts, the ease with which these groups used ChatGPT to create professional-sounding reports and fake news stories remains a major concern for global security. For now, the company says it is committed to investigating these activities and preventing its tools from being used to manipulate public opinion or cause real-world harm.