- Filed
- September 10, 2026, 07:20 UTC
- Actor
- Russia-origin operators (OpenAI attribution via Russian-language prompting, VPN circumvention of geo-blocks, Slavic translation artifacts; consistent with prior Russia-linked IO patterns)
- Target
- English- and other-language audiences on Western social platforms, public perceptions of Russian vs. Western sovereignty, and legitimacy of pro-Kremlin geopolitical framing
- Confidence
- documented Primary documents, court records, official disclosures
- Techniques
- Synthetic mediaNarrative launderingCard stackingAstroturfing
- Channels
- ChatGPT-generated social posts and comments
- fake think-tank website with plagiarized academic content
- Burke Sovereignty Index reports
- multi-platform distribution (X, LinkedIn, Facebook, Substack, Telegram)
- VPN access and language-concealment prompts
- inauthentic everyday-user and branded accounts
Summary
OpenAI disrupted a cluster of ChatGPT accounts originating in Russia that generated 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 plagiarized and misattributed academic articles, a proprietary Burke Sovereignty Index scoring Russia highly while criticizing Western states, and efforts to conceal Russian origins via VPN access and language-hiding prompts. Content was pushed to X, LinkedIn, Facebook, Substack and Telegram; reach remained limited but infrastructure was elaborate.
Analysis
Primary documentation is OpenAI’s 25 August 2026 public report on the disruption, corroborated by contemporaneous coverage from the Council on Foreign Relations, Le Monde, Der Spiegel references, and archival captures of the IBI site. Operators accessed ChatGPT via VPNs despite Russia geo-blocks, prompted in Russian for English (and other) social posts that promoted IBI articles while instructing the model to hide linguistic traces of non-native origin. The core asset was the IBI website, presented as an Israel-based expert community listing prominent names including Francis Fukuyama and Noam Chomsky among affiliates (none of whom had affiliation). Of a sample of 36 articles, 34 were copied from legitimate sources and often misattributed; residual Slavic phrasing (e.g., “Svetofor coalition” for Germany’s Ampelkoalition) indicated machine translation from a Slavic language. The proprietary Burke Sovereignty Index assigned high scores to Russia across political, economic, military and other dimensions while framing Western states and the EU as sovereignty-deficient, dependent, or in decline. Parallel Telegram channels and replies on authentic Substack posts sought amplification.
The mechanism is narrative laundering through synthetic institutional form. AI lowers the cost of generating polished, platform-native promotional text; plagiarism of real scholarship supplies surface legitimacy while false attributions and an invented index create the appearance of independent expertise. Card-stacking is evident in the selective scoring and polemic framing of the index reports. Astroturf effects arise when branded and everyday-looking accounts circulate the material as organic expert discourse rather than coordinated advocacy. Concealment tactics—VPN, language-hiding instructions, Israel branding—constitute operational security aimed at defeating attribution.
This qualifies as an influence operation because authenticity is artificially constructed (non-existent institute with borrowed reputations, AI-authored promotion, plagiarized content presented as original research, obscured Russian origin), coordination is present across generation and distribution assets, and the design goal is to shape perceptions of relative sovereignty in ways favorable to Russian strategic narratives. Technique families are scored independently of the actor; identical patterns of synthetic think-tank creation, academic plagiarism for legitimacy, proprietary indices, and AI-assisted multi-platform promotion have been observed across state and non-state operators. Limited documented audience reach does not negate the infrastructural investment or the potential for downstream citation and narrative seeding.
Literacy counter: treat newly appearing institutes, indices or expert communities that lack verifiable staff, original peer-reviewed output, transparent funding or physical presence as provisional, especially when their framing clusters tightly around a single state’s preferred geopolitical narrative. Cross-check claimed affiliations and article authorship against primary academic sources and institutional records. Prefer original statistical agencies, established think tanks with disclosed governance, and multi-source verification over any single branded index optimized for social amplification. Platforms and model providers should surface residual translation artifacts, bulk generation patterns and institutional-provenance signals earlier in the content lifecycle.
Evidence
- OpenAI: Disrupting a new covert influence campaign from Russia (25 Aug 2026)
- Council on Foreign Relations: The Influence Operation That Ran on Borrowed Reputations (3 Sep 2026)
- Ctech / Calcalist: OpenAI exposes Russian campaign promoting fake Israeli research institute (26 Aug 2026)
- IBI Sovereignty Index sample page (archived)