OpenAI bans Russia-linked ChatGPT accounts used in covert disinformation campaign

OpenAI has blocked a group of ChatGPT accounts traced to Russia that the company says were used to support a covert online influence operation and spread disinformation.

The company says it uncovered the network after investigating multiple AI-generated social posts, a probe that revealed "a much broader influence operation."

OpenAI said the operation "was built around a website with copied and misattributed academic papers, an index of sovereignty, which portrayed Russia in a favorable light, as well as efforts to hide the Russian origin of the operators." The materials were used to give apparent credibility to the campaign's narratives.

AI used to create apparent authority

According to OpenAI, operators fed Russian-language instructions into ChatGPT to generate comments for social platforms; those posts were then published on Substack, Telegram, X, Facebook and LinkedIn. The company said ChatGPT was instructed to remove linguistic traces that would indicate a Russian origin, and that virtual private networks (VPNs) were used to mask access.

Much of the generated content promoted an organization called the International Burke Institute. OpenAI said the institute's site included articles copied from real academic papers, sometimes with false attribution of authorship. The campaign also produced an "index of sovereignty" that praised Russia and disparaged Western countries.

OpenAI acknowledged that the immediate reach of the operation appeared limited, but warned the deeper concern was the infrastructure the campaign created: posts and websites that mimic legitimate institutions, cite alleged experts and republish academic material to establish credibility over time.

In February OpenAI said it had closed ChatGPT accounts linked to Rajber, a pro‑Russian media company that the British government describes as "partially coordinated by the presidential administration," funded by the Russian state corporation Rostec and cooperating with members of Russian intelligence services. The recent action shows how actors can combine online platforms and AI tools to build networks of seemingly authoritative content.

Photo: Jakub Porzycki/NurPhoto / Shutterstock Editorial / Profimedi