OpenAI Restricts ChatGPT Access to Russian Propagandists Who Developed Psychological Operation Plans

OpenAI Blocks Russian Propaganda Network Using ChatGPT for Disinformation Campaigns

OpenAI has taken decisive action against a Russian propaganda network known as “Rybar” by blocking multiple ChatGPT accounts linked to the organization. The accounts were identified during a counter-intelligence operation dubbed “Fish Food,” which revealed an extensive operation leveraging AI tools to generate content for propaganda purposes.

According to a recent OpenAI report, the propaganda network utilized ChatGPT to create posts and comments that were subsequently distributed across various social media platforms. The content appeared under the official “Rybar” brand as well as through seemingly unrelated accounts designed to appear as authentic users from different regions around the world.

“We banned a cluster of ChatGPT accounts that were linked to the Russian-origin ‘Rybar’ network,” OpenAI stated in their report. “This cluster translated and generated content that was posted on ‘Rybar’ social media accounts, but it also appears to have served as a content farm for a wider network of accounts on X and Telegram that bore no overt relationship to the ‘Rybar’ group.”

The investigation found that the propaganda operatives frequently used ChatGPT to mass-produce batches of short social media comments. These comments were then disseminated by accounts on X and Telegram strategically designed to appear as though they originated from diverse geographical locations, creating an illusion of widespread organic engagement.

While the blocked users communicated in Russian, they produced multilingual propaganda materials in several languages including Spanish and English. The report detailed that these materials consistently “praised Russia and its allies (such as Belarus), criticised Ukraine and accused Western countries of foreign interference.”

The operation’s sophistication extended beyond text generation. Rybar operatives also utilized OpenAI’s advanced video generation tool, Sora, to create promotional videos supporting their messaging campaigns. This represents one of the first documented cases of AI-generated video being weaponized for state-backed propaganda efforts.

Perhaps most concerning was the discovery that one of the blocked accounts had requested creation of detailed information and psychological operation plans specifically designed to interfere with political processes and elections across several African nations. Evidence suggests the propagandists were establishing a network of agents and planning large-scale events and protests targeting the Democratic Republic of the Congo, Cameroon, Burundi, and Madagascar.

The scale of these operations appears significant, with the report noting that the budget for the largest propaganda operation in Africa amounted to US$600,000. This aligns with broader concerns about Russian influence operations on the continent, which have intensified in recent years as Moscow seeks to expand its geopolitical influence.

The “Rybar” network has been on Western intelligence agencies’ radar for some time. The OpenAI report noted that two years ago, the United States government offered a reward of up to US$10 million for information regarding the “Rybar” project, underscoring its significance as a threat to information integrity.

This development comes amid growing concerns about the misuse of generative AI technologies for disinformation campaigns. As AI tools become more sophisticated and accessible, their potential for weaponization in information warfare continues to present significant challenges for technology companies and national security agencies alike.

OpenAI’s action represents one of the most high-profile interventions by an AI company against state-affiliated propaganda networks, highlighting both the evolving nature of digital influence operations and the increasing responsibilities falling to private technology companies in combating them.