Artificial intelligence developer OpenAI revealed on Tuesday that it has disrupted a sophisticated, covert Russian influence operation that leveraged ChatGPT accounts to establish and promote a fictitious research institute falsely claiming to be based in Israel.
The Russian operators utilized Virtual Private Networks (VPNs) to circumvent geographic platform restrictions, directing the language models to generate social media posts and replies in English and German while deliberately disguising Russian linguistic traces.
The propaganda network was systematically deployed across X, Facebook, LinkedIn, Substack, and Telegram.
Fake Research Institute with Israeli Address
At the heart of the campaign stood an entity branded as the “International Burke Institute” (IBI), whose domain was registered in February 2025 boasting a physical address in Israel.
The website presented itself as a prestigious “expert community,” deceptively listing the names of world-renowned scholars and public intellectuals.
However, an audit conducted by OpenAI on a sample of 36 articles published by the institute revealed that 34 had been plagiarized directly from legitimate academic publications and falsely attributed, alongside poorly structured texts translated via automated tools from Slavic languages.
The sham institute served as a launchpad to publish a fabricated “Sovereignty Index” designed to portray the Russian Federation in an overwhelmingly positive light while systematically smearing Western democracies opposed to Moscow’s invasion of Ukraine.
Fabricated reports attacked the United States, France, Germany, and the European Union to erode public trust in Western institutions.
Concurrently, another operator generated AI profile graphics and branding for dozens of regional Telegram channels pushing masked pro-Kremlin narratives.
The Israeli Connection and Exposure of the Infrastructure
A particularly concerning finding in the threat intelligence report indicates that a small number of real individuals residing in Israel ostensibly represented the IBI institute by submitting academic papers and attending professional conferences.
OpenAI stated that it cannot definitively establish the precise relationship between those Israeli individuals and the network operators in Russia, or whether they acted in good faith after being deceived by the fake institute’s legitimate façade.
Despite the elaborate digital infrastructure, OpenAI assesses that the campaign’s organic reach and viral impact remained relatively constrained, with most social media posts generating negligible engagement and Telegram channels drawing limited subscriber pools.
Nevertheless, the company warned that the operation represents an alarming escalation in how state-linked threat actors harness generative AI to fabricate scientific credibility, obfuscate state propaganda sources, and undermine global information integrity.