"were being used to promote the International Burke Institute (IBI), a self-described 'expert community' based in Israel," OpenAI said.
OpenAI’s takedown and what it found
OpenAI announced that it banned a cluster of Russian ChatGPT accounts after an investigation began with AI-generated social media posts and expanded into a broader influence operation. According to the company, the accounts used VPNs to bypass access restrictions and relied on ChatGPT to generate social media posts and comments that were then shared across Substack, Telegram, X, Facebook and LinkedIn.
OpenAI described the campaign as distinct from already documented Russia-linked influence operations and emphasized that, while the audiences reached were "relatively small," the operation had built a notable infrastructure that supported its messaging.
The International Burke Institute and the Burke (Sovereignty) Index
The accounts promoted a site called the International Burke Institute (IBI) — associated with the domain ibi[.]institute — which the company says was registered in February 2025 and presents itself as a community of experts on economics, politics, sociology and international relations. Central to the site's output was a so-called Sovereignty Index, also called the Burke Index.
The IBI website explains the Burke Index is calculated from seven equal-weight indicators: political, economic, technological, information, cultural, cognitive, and military sovereignty. "The sum of 7 indices in the range of 100-700 forms the final Cumulative Sovereignty Index," the site states. The published index ranks the United States first at 650.9, China second at 649.1, Switzerland third at 610.7, and Russia fourth at 601.4.
A report on Russia hosted by the site describes Russia in 2025 as "a state with a high level of sovereignty and record military and economic self-sufficiency" while noting limits in civilian economic flexibility and dependence on "internal administrative capital." The report argues the country has become "one of the three world powers with full digital and military nuclear sovereignty" and lists demographic aging and innovation constraints as ongoing challenges.
Audit-ready is a season. It shouldn't be.
Evidence in spreadsheets, controls drifting between audits, frameworks multiplying on flat headcount. Nubivance runs continuous compliance on Rapid7 Cyber GRC - SOC 2, HIPAA, ISO 27001, PCI, CMMC.
End the scrambleTechniques: VPNs, AI prompts, translation and misattribution
OpenAI said operators sought to obscure their Russian origins. The majority of the generated content was in English, and operators instructed ChatGPT to conceal linguistic clues that could reveal Russian provenance. In some cases articles on the IBI site appeared copied from academic writings, sometimes with false attribution; other pieces looked like drafts produced by a Slavic speaker and then machine translated into English. OpenAI noted there is no evidence articles on the site were created using ChatGPT.
One of the ChatGPT users also used the tool to create a profile picture for a Telegram channel named "American Observer" and repeatedly asked for Russian-language summaries of that channel’s activity, according to OpenAI.
Reach, channels and the operation’s infrastructure
OpenAI assessed the campaign reached "relatively small audiences," although Telegram channels attracted more users and amassed about 10,000–20,000 followers each. The company stressed that the significance of the operation was less the immediate audience than the infrastructure: a credible-appearing institution with purported experts, republished academic work and a proprietary risk index that could be used to amplify favored narratives.
"While the actors only used ChatGPT to produce isolated promotional posts, those posts pointed to an otherwise credible-appearing institution," OpenAI said, arguing the combination of AI-produced content and a layered asset — website, reports, indexes and social channels — is what made the operation notable.
What this means for technologists, policymakers, and the public
- Technologists and security teams: The case shows how AI can be a supporting tool inside a larger information operation, used alongside VPNs, machine translation and reused academic material to manufacture perceived authority. OpenAI highlighted that the actors used ChatGPT to generate isolated promotional posts that linked to a broader infrastructure.
- Policymakers and regulators: The operation underlines how products and platforms can be leveraged to create institutions that project credibility — including through indices and republished reports — and how provenance can be deliberately obscured. OpenAI described the campaign as distinct from previously documented influence operations, signaling a different model of operation to consider.
- End users and the public: Even campaigns that reach "relatively small audiences" can leave behind assets — websites, indexes, channels — that appear authoritative. OpenAI framed the episode as an illustration that AI can both support disinformation efforts and, when used in this way, help expose them: "This illustrates how influence actors can use AI as a supporting tool within a broader effort to manufacture authority ... It also illustrates how their supporting use of AI can lead to the broader operation being exposed."
OpenAI’s announcement leaves a clear throughline: the tool was one component in a deliberate, multi-part effort built around a branded institution and a proprietary index. Whether similar operations scale or evolve remains an open question in the company’s account, but the finding stresses that influence operations increasingly pair synthetic content with curated, outward-facing infrastructure.
