Russia-origin operators use ChatGPT to promote…

Filed

September 10, 2026, 02:15 UTC

Actor

Russia-origin covert influence operators (OpenAI attribution; VPN access, Russian-language prompting, Slavic translation artifacts; linked to promotion of pro-Kremlin narratives)

Target

English- and German-speaking Western audiences on social platforms; secondary influence on AI retrieval and public perception of national sovereignty rankings

Confidence

documented Primary documents, court records, official disclosures

Techniques

Synthetic media Narrative laundering Astroturfing Card stacking Manufactured consensus

Channels

ChatGPT-generated social posts and replies
fake think-tank website with misattributed academic content
Burke Sovereignty Index
VPN-obfuscated access
Telegram channels and multi-platform amplification

Summary

OpenAI disrupted a cluster of ChatGPT accounts very likely originating in Russia that generated English-language social media posts promoting the International Burke Institute (IBI), a self-described expert community claiming an Israel base. The IBI website, registered February 2025, hosted a Burke Sovereignty Index favoring Russia while criticizing Western states, plus dozens of academic articles that OpenAI found were largely copied and often misattributed. Operators used VPNs, prompted in Russian, and instructed the model to conceal linguistic tells; content appeared on X, LinkedIn, Facebook, Substack and Telegram.

Analysis

Primary documentation is OpenAI’s 25 August 2026 report detailing the ban of a cluster of ChatGPT accounts that originated in Russia. Operators prompted in Russian, employed VPNs to bypass geographic restrictions, and explicitly instructed the model to strip linguistic markers of Russian origin. The generated English (and some German) posts promoted the International Burke Institute website and its content across Substack, Telegram, X, Facebook and LinkedIn. Parallel non-AI activity centered on the IBI site itself, registered in February 2025 and presented as an Israel-based expert community listing prominent scholars including Francis Fukuyama and Noam Chomsky. OpenAI sampled 36 articles linked to the site’s purported experts (September 2025–May 2026) and found 34 were copied from external sources, frequently with incorrect author attribution—for example, a Cambridge University Press piece on the China-Pakistan Economic Corridor reassigned to a Nottingham professor, and Migration Policy Institute material credited to an Australian food-chemistry academic whose photo was reused.

The Burke Sovereignty Index supplied the core narrative payload: comparative rankings that elevated Russia while criticizing France, Germany, the United States, the EU and other states critical of the war in Ukraine. Sample report language is polemical rather than analytic (e.g., Macronism characterized as opening France “like a safe with historical capital”). Card-stacking is therefore structural: selective metrics and framing privilege the desired geopolitical conclusion. Narrative-laundering occurs through the institutional camouflage of a faux think tank that borrows real academic reputations and host-country branding (Israel) to lend apparent independence. Astroturf and manufactured-consensus effects arise from the coordinated appearance of multiple quasi-organic accounts posting and replying in ways that direct attention to the same source. Synthetic-media techniques are limited to the social-layer content generation; the website’s longer texts were largely plagiarized rather than LLM-authored.

This constitutes an influence operation because authenticity is artificially constructed (fictitious institute, misattributed expertise, concealed operator nationality), origin is deliberately obscured, and the design goal is to shape perceptions of national legitimacy and Western policy toward Russia. Technique families are scored independently of the actor; identical patterns of institutional mimicry, reputation theft, and LLM-assisted distribution have been observed across commercial, party and state actors. OpenAI assessed reach as relatively small, yet the infrastructure—VPN access, multi-platform seeding, reusable index product—is scalable and illustrates how low-cost generative tools lower the barrier to constructing entire pseudo-institutions.

Literacy counter: treat any newly appearing “institute,” “expert community” or proprietary index that lacks verifiable staff, physical presence, peer-reviewed output or transparent funding as provisional, especially when its rankings cluster tightly around a single geopolitical narrative. Cross-check claimed authors against primary publications, institutional directories and contemporaneous records. Prefer original academic sources, official statistics and multi-source journalism over any single branded report optimized for social amplification. Platforms and model providers should surface provenance signals, foreign-origin indicators and coordination patterns for synthetic institutional content before it enters retrieval corpora.

Russia-origin operators use ChatGPT to promote fabricated Israel-based International Burke Institute and pro-Kremlin sovereignty index

Summary

Analysis

Evidence