OpenAI Dismantles Russian Influence Operation Using ChatGPT to Spread Western Disinformation

OpenAI has disrupted a clandestine online influence campaign originating in Russia that leveraged ChatGPT to generate propaganda and manage cross-platform disinformation operations aimed at undermining Western support for Ukraine, as first reported by TechRadar Pro.

According to findings released by OpenAI, threat actors utilized automated accounts to build credibility for a pseudo-academic organization styled as the International Burke Institute (IBI). Although registered to an address in Israel, the enterprise operated primarily to amplify Kremlin-aligned political messaging and portray Western governments unfavorably relative to Moscow.

The actors behind the network implemented specific operational security measures within ChatGPT to obscure their Russian roots. Prompts submitted to the artificial intelligence model explicitly instructed it to eliminate any linguistic patterns or stylistic markers that would reveal the initial queries were written in Russian.

Despite those precautions, technical indicators of Slavic language origin persisted in published materials. For instance, text on the International Burke Institute portal translated Germany’s ruling political alliance as the "Svetofor coalition," drawing directly from the Russian word for traffic light rather than standard international terminology, indicating that underlying content was drafted in a Slavic language prior to translation.

Beyond web hosting, the campaign deployed automated accounts across platforms including X, LinkedIn, Facebook, Substack, and Telegram. Operators relied on ChatGPT to create matching avatar images, produce synthetic visual media, generate accompanying text posts, and automatically post replies to real user comments on Substack. The operators frequently asked the AI system to output page performance metrics in Russian.

To establish an aura of legitimacy, the fake institute published a scoring metric called the Burke Sovereignty Index, which evaluated nations across political, cognitive, economic, technological, cultural, and military factors. The system calculated skewed results designed to rank Western democracies below Russia, assigning Moscow top military ratings while producing illogical outputs such as placing Vatican City ahead of Spain.

The organization further inflated its standing by publishing directories of renowned academics and political figures without their knowledge or consent. The site listed scholars Francis Fukuyama and Noam Chomsky alongside former senior U.S. government officials Mike Pompeo and Joseph Nye, while also claiming associations with deceased individuals including Shlomo Avineri and Jiang Ping. Additionally, legitimate research papers were copied and credited to unrelated authors to hide original sources while populating the platform with authentic content.

OpenAI assessed the overall audience reach of the campaign as modest, noting that associated Telegram channels accumulated limited followings ranging between 10,000 and 20,000 subscribers. Nevertheless, the extensive underlying framework highlighted an escalating effort by covert actors to integrate commercial AI tools into complex disinformation networks.