Cyber & Technology — 2026-08-26

OpenAI Disrupts Russian Covert Influence Campaign Operating Fake Israeli Think Tank to Manipulate AI Search Results

BLUFRussia's use of AI to promote fabricated authority structures rather than generate content directly undercuts platform detection models built around synthetic-text signatures.

OpenAI banned a cluster of ChatGPT accounts, originating in Russia and accessed via VPN, that generated social media content promoting the International Burke Institute (IBI), a self-described Israel-based "expert community" registered in February 2025 1. OpenAI found that 34 of 36 sampled IBI articles were copied from other sources with false or misattributed authorship, and that the site published a "sovereignty index" scoring countries in terms favorable to Russia 1. Operators generated English-language posts across Substack, Telegram, X, Facebook and LinkedIn while instructing ChatGPT to conceal linguistic signs of Russian origin, and a second operator built logos for roughly a dozen additional Telegram channels targeting Germany, the United States, France, Poland and Turkey, including German-language content criticizing Ukraine, the EU and the German government 1. OpenAI rated the campaign Category Three of six on the Brookings Breakout Scale and assessed its reach as limited, with its Telegram channels drawing roughly 10,000 to 20,000 followers each; the action follows a February action against accounts linked to the pro-Russia outlet Rybar 12.

Analysis
The disruption removes one asset while leaving the underlying method intact: a stocked-article website and "sovereignty index" built to project independent authority, with ChatGPT used only for peripheral promotion rather than content generation, a design choice that makes future clusters harder to flag through content-based detection alone. Taken with February's action against Rybar-linked accounts, this points to a standing Russian pattern of testing chatbot platforms for auxiliary influence functions rather than isolated misuse, though the individuals who represented the institute at conferences and in article submissions may be witting participants lending it genuine credibility rather than unwitting cover. OpenAI's disclosure is the sole primary account of the platform-side evidence; outlets covering the case add no independent sourcing. Current reporting does not address whether operators rebuild a comparable front.

4 sources