Full Report
OpenAI said it banned a network of ChatGPT accounts linked to the pro-Kremlin media outlet Rybar that used artificial intelligence to generate social media content and draft proposals for covert influence operations targeting Africa. The campaign, dubbed “Fish Food,” generated large batches of multilingual content that were later posted across Telegram and X. At least…
Analysis Summary
# Threat Actor: Rybar (Campaign: “Fish Food”)
## Attribution & Identity
* **Actor Identification:** Rybar, a prominent pro-Kremlin media outlet and influential Russian military blogging entity.
* **Aliases:** Linked to broader Russian influence operations; the specific ChatGPT-based campaign is dubbed **"Fish Food"** by OpenAI.
* **Known Associations:** Historically associated with the Russian Ministry of Defense and formerly linked to the Wagner Group’s propaganda ecosystem.
## Activity Summary
In early 2026, OpenAI identified and dismantled a network of ChatGPT accounts utilized by Rybar to facilitate covert influence operations (IO). The campaign, "Fish Food," focused on automating the production of disinformation and propaganda. The actor used Generative AI to create high volumes of multilingual social media content and authored detailed proposals for steering public opinion in Africa.
## Tactics, Techniques & Procedures
* **AI-Enhanced Content Generation:** Leveraging Large Language Models (LLMs) to function as a "content farm" for generating social media posts.
* **Multilingual Scripting:** Developing content in multiple languages to increase the reach and resonance of propaganda across diverse regional audiences.
* **Operational Planning:** Using ChatGPT to draft strategic proposals for covert influence operations.
* **Cross-Platform Distribution:** Disseminating AI-generated material via Telegram and X (formerly Twitter) to mimic organic engagement.
* **Infrastructure Obfuscation:** Using accounts that likely originated in Russia to interact with Western AI services to bypass traditional content moderation.
## Targeting
* **Sectors:** Media, Government, and Public Opinion (Information Environment).
* **Geography:** Primarily targeting Africa; infrastructure and accounts originated in Russia.
* **Victims:** General social media users on Telegram and X; regional African audiences targeted by pro-Kremlin revisionist narratives.
## Tools & Infrastructure
* **AI Tools:** OpenAI’s ChatGPT (various accounts now banned).
* **Distribution Channels:**
* t[.]me (Telegram)
* x[.]com (X/Twitter)
* **Infrastructure:** Accounts assessed by OpenAI to be originating from Russian geography.
## Implications
The "Fish Food" campaign demonstrates the evolving "industrialization" of influence operations. By integrating Generative AI, threat actors like Rybar can significantly lower the cost of producing high-quality, multilingual propaganda while increasing the velocity of information warfare. The use of LLMs to draft covert operational plans suggests that Russian influence actors are moving beyond message generation and into AI-assisted strategic planning.
## Mitigations
* **Platform Monitoring:** Continued collaboration between AI providers (OpenAI, Anthropic, etc.) and social media platforms to identify and link "content farms" to state-backed actors.
* **Behavioral Analysis:** Defense focus on identifying bot-like distribution patterns on X and Telegram, as AI content is increasingly indistinguishable from human writing.
* **Public Awareness:** Educational initiatives regarding the use of AI in foreign influence campaigns, particularly in targeted regions like Africa.
* **API Restrictions:** Implementation of stricter verification for high-volume accounts or those exhibiting patterns consistent with known state-sponsored propaganda outlets.