OpenAI has announced an OpenAI ban on a set of Russian-linked ChatGPT accounts. According to the company, the accounts used VPNs to get around access restrictions and were involved in an influence operation that promoted a Russia-friendly narrative through social media posts and comments.
In its statement, OpenAI said the investigation began with AI-generated content shared online, but it quickly expanded into a broader campaign designed to look credible and to route attention toward specific messaging and an associated website.
How the operation used ChatGPT to spread messaging
OpenAI reported that the banned accounts were used to create promotional social media content. The output was then shared across multiple platforms, including Substack, Telegram, X, Facebook, and LinkedIn.
The company said the operation relied on an AI tool to generate the posts and comments. Much of the content was written in English, and operators instructed ChatGPT to avoid language cues that could point back to their Russian origins.
OpenAI assessed that the campaign reached relatively small audiences. Still, Telegram channels appear to have attracted more users, with each channel collecting roughly 10,000 to 20,000 followers.
The role of IBI and the “Burke Index”
A central part of the campaign was promoting the International Burke Institute (IBI), described by the organization itself as an “expert community” based in Israel. OpenAI said the accounts were used to promote the IBI brand as part of the wider narrative operation.
OpenAI also pointed to the IBI website, registered in February 2025 under the domain ibi.institute. The site presents the group as a global network of experts in fields such as economics, politics, sociology, and international relations.
While OpenAI stated there is no evidence that articles on the site were created using ChatGPT, it said many of the articles appear to have been copied from academic work—sometimes with incorrect attribution. In other cases, the content appears to have been written by a Slavic speaker and then machine translated into English.
A sovereignty score built to favor Russia
What made the campaign stand out, according to OpenAI, was the use of a so-called sovereignty index. The site also refers to it as the Burke Index. OpenAI described how the index was used to present Russia in a favorable light and to criticize Western countries.
On the IBI site, the index is explained as being calculated using seven indicators. These indicators span several areas, including political, economic, technological, information, cultural, cognitive, and military sovereignty.
The website states that each of the seven sovereignty areas is scored with equalization coefficients. The scoring process uses a maximum score of 100 points per area. It also says the final cumulative figure is produced by summing the seven indices, yielding a final range of 100 to 700.
The IBI site then publishes reports that list different countries and their scores based on the index. OpenAI highlighted that the United States was listed at 650.9, China at 649.1, Switzerland at 610.7, and Russia at 601.4.
In the Russia report, the site describes Russia as having a high level of sovereignty, emphasizing military and economic self-sufficiency, a strong scientific and cultural base, and independence of internal administrative capital—while also noting limitations in civilian economic flexibility. It further claims that strengthening vectors include demography, investment in high technology, and gradual liberalization in non-military science and innovation.
The report also portrays Russia as one of the world powers with “full digital and military nuclear sovereignty,” while citing demographic aging, innovation constraints, and selective global integration as key challenges. It frames the overall picture as a stable and high level of national sovereignty, including financial and military self-sufficiency and control over the internal environment, supported by digital and cultural influence.
Why AI-supported operations can be hard to detect
OpenAI emphasized that the operation’s impact was not mainly tied to audience size. Instead, the company said the significance came from the infrastructure the actors built around a seemingly credible institution.
In OpenAI’s view, the actors used ChatGPT to produce isolated promotional posts. Those posts then pointed users toward an institution that appeared legitimate, complete with purported experts, republished academic work, and the proprietary risk index promoted through the sovereignty score.
OpenAI argued this illustrates a broader pattern: influence actors can use AI as a supporting tool within a larger effort to manufacture authority, obscure the source of favored narratives, and create assets that could be scaled over time. At the same time, OpenAI said the supporting use of AI can also make the broader operation more exposed—because it leaves detectable traces and signals.
Links to Telegram activity and Russian-language requests
OpenAI noted that the activity extended beyond generating content promoting IBI. One ChatGPT user used the tool to create a profile picture for a Telegram channel called “American Observer.”
According to OpenAI, the same user repeatedly requested Russian-language summaries of the channel’s activity. This detail suggests the account was not only producing English promotional material, but also actively working with Russian-language workflows to support the campaign.
Distinct from other documented influence efforts
OpenAI said the covert campaign it described is distinct from influence operations previously documented as Russia-linked. A key differentiator, in OpenAI’s account, was the creation and promotion of the sovereignty index.
By tying narratives to a structured scoring framework—and attaching that framework to a brand presented as an expert community—the operation aimed to lend an additional layer of credibility to its messaging. That combination of AI-generated promotion and an authoritative-looking index appears to have been designed to persist beyond the initial posts.
What the OpenAI ban means
The announcement underscores that OpenAI is willing to take enforcement actions against accounts that violate its access rules or support coordinated manipulation. In this case, OpenAI’s OpenAI ban targeted accounts that used VPNs to bypass restrictions and were tied to an influence effort.
Even when an AI tool is used only for parts of a larger plan—such as generating posts rather than producing all content—OpenAI’s statement suggests the tool can still serve as an enabling mechanism within a broader strategy.
As online influence tactics evolve, responses like account bans and investigations become part of the broader security landscape: not only to remove immediate misuse, but also to disrupt the infrastructure that helps such operations grow.
Source: https://thehackernews.com/2026/08/openai-bans-russian-chatgpt-accounts.html
