OpenAI Disrupts Russian Influence Operation Using ChatGPT

OpenAI has banned a cluster of ChatGPT accounts linked to a Russian disinformation operation that used the AI to generate social media posts. The campaign promoted the International Burke Institute (IBI), a fictional Israeli think tank that published content critical of Western nations and supportive of Russia. The operation targeted platforms including X, Telegram, Substack, Facebook, and LinkedIn. To bypass OpenAI's ban on access from Russia, the actors used VPNs. OpenAI reported that the prompts were written in Russian and included instructions to hide linguistic clues that would reveal the origin of the text. However, OpenAI noted that some attempts to mask the translation failed, such as the use of the term Svetofor coalition to describe a German political group. The IBI website featured a sovereignty index that praised Russia and criticized countries opposing its invasion of Ukraine, including France and Germany. OpenAI found that 34 out of 36 sampled articles on the IBI site were copied from other sources, sometimes with incorrect author attributions. None of the articles on the IBI site were generated by ChatGPT. OpenAI stated that the immediate impact of the campaign was limited due to low view counts and subscriber numbers, though it noted the elaborate construction of the operation distinguished it from previous disruptions.