OpenAI bans Russian ChatGPT accounts used in covert…

In May 2024, OpenAI identified and terminated a network of accounts originating from Russia that were actively facilitating a covert online influence operation. The campaign leveraged AI-generated content to disseminate misinformation across multiple social media platforms, including X, Telegram, Facebook, LinkedIn, and Substack.

What happened

The operation utilized ChatGPT to generate social media commentary, specifically instructing the model to hide linguistic markers that might reveal the operators' Russian origins. By utilizing VPNs to circumvent geographic restrictions, the actors accessed the platform to build a deceptive infrastructure centered around a fabricated entity known as the International Burke Institute (IBI).

This campaign was designed to manufacture institutional authority by populating the IBI website with copied academic works, often featuring false attributions, and a "sovereignty index" that portrayed Russia favorably while disparaging Western nations. The operators used the AI to create content that pointed toward these fabricated assets, effectively attempting to establish a veneer of credibility through purported experts and proprietary research indices.

OpenAI noted that while the immediate reach of the operation appeared limited, the underlying infrastructure was designed for scalability. The discovery highlights how influence actors are increasingly using generative AI as a supporting tool to obscure the sources of favored narratives and manufacture false credibility. This incident follows a February 2024 enforcement action in which the company shut down accounts linked to the pro-Russia media company Rybar. OpenAI emphasized that the significance of this operation lies in the infrastructure built to deceive the public, rather than the immediate audience captured by the misinformation.