Disrupting a new covert influence campaign from Russia

OpenAI has disrupted a covert influence campaign originating from Russia that utilized ChatGPT accounts to manipulate public opinion. According to OpenAI, the operation centered around promoting the International Burke Institute, a self-described expert community that falsely claimed to be based in Israel. The website featured a sovereignty index designed to praise Russia while criticizing Western nations, the European Union, and countries critical of the war in Ukraine, such as Germany and France. To execute the campaign, operators used VPNs to access OpenAI platforms from Russia and prompted in Russian to generate social media comments in English across platforms like Substack, Telegram, X, Facebook, and LinkedIn. The operators deliberately instructed ChatGPT to hide linguistic clues indicating their Russian origin. While the IBI website content itself was not generated by OpenAI models, investigations revealed it was stocked with copied academic articles, often with misattributed authors or machine-translated text containing Slavic phrasing anomalies. In addition to promoting the IBI brand, operators generated German-language posts critical of Ukraine and the EU for a Telegram channel and created profile assets for various regional channels targeting the United States, France, Poland, and Türkiye. OpenAI stated that this elaborate construction distinguishes the campaign from other Russia-linked operations disrupted since the start of the war in Ukraine, underscoring ongoing challenges in detecting deceptive AI-assisted influence operations.