OpenAI bans Russian ChatGPT accounts posing as a fake Israeli think tank — used VPNs to push pro-Kremlin narratives and steal academic papers for its website

On August 25, OpenAI banned a network of ChatGPT accounts involved in a disinformation campaign centered around a fake Israeli think tank. The think tank, known as the International Burke Institute (IBI), fabricated research and stole academic papers from other authors to promote pro-Kremlin narratives. The operators behind the IBI used VPNs to access OpenAI's models from Russia, where the platform is blocked.

They created 34 articles from real academic works, misattributing them to notable figures such as Francis Fukuyama and Noam Chomsky.

The IBI website was registered in February of the previous year, while the articles, country reports, and index were not produced with OpenAI's models. Instead, ChatGPT was used to rewrite the content to remove any indicators that it originated from Russian speakers, but the AI-generated material contained telltale signs of artificial intelligence.

One article even mislabeled Germany's traffic-light coalition as the "Svetofor coalition," a direct translation of the Russian term for traffic light, which would not be used in either English or German.

OpenAI's intelligence team stated that the campaign ranked at the lower end of Category Three on their six-category Breakout Scale, indicating low view counts and limited scope. However, the significance of the operation lies in the infrastructure built by the operators, which they had previously used to amplify other narratives. OpenAI's ban was a response to operators utilizing VPNs to access their services for this specific disinformation campaign.