OpenAI has dismantled a sophisticated influence network tied to the Russian-aligned “Rybar” brand, dubbing the operation “Fish Food” for its reliance on AI-generated content farms. The ban targets ChatGPT accounts that produced multilingual propaganda in Russian, English, and Spanish, feeding both official Rybar channels and seemingly independent social media profiles. This isn’t just about spam; it represents a shift toward industrial-scale disinformation where AI handles the heavy lifting of content creation while human operators manage distribution.
The scale of ambition here is startling. Investigators found prompts detailing commercial plans for covert interference campaigns across Africa, with budgets reaching $600,000 annually for the most aggressive projects. These proposals weren’t vague musings. They included specific strategies for inflaming protests in Madagascar, manipulating electoral processes in Burundi and Cameroon, and establishing a network of local agents in the Democratic Republic of Congo. One user even requested edits for a deployed election interference team, blending online amplification with on-the-ground activism.
Data from the operation reveals a crucial insight about viral reach. When a single prompt generated seven tweets posted by different accounts, viewership ranged wildly from 57 to over 150,000 views. The variance correlated directly with follower counts, not the quality or origin of the AI text. This suggests that audience size, not algorithmic preference for synthetic content, drives visibility. It undermines the fear that AI posts inherently trick platforms into wider distribution; instead, they simply amplify existing reach.
Beyond election meddling, the network promoted “REST Media,” an outlet linked to Rybar, by generating comments that drove traffic to articles accusing Germany of interference in Moldova. OpenAI’s detection relied on matching exact language strings from ChatGPT outputs to posts on Telegram and X. While the technical mechanism for posting remains unclear, the strategic intent is plain. As AI lowers the cost of producing persuasive text, state-aligned actors are scaling up operations that blend digital noise with real-world political disruption.
