The Website Looked Like a Real Institution
It had experts, research papers, a physical address, and its own global ranking system.
It was also the center of what OpenAI describes as a newly uncovered covert influence operation originating in Russia.
The International Burke Institute presented itself as an expert community based in Israel. Behind that professional exterior, OpenAI found copied academic work, false attribution, coordinated social accounts, and ChatGPT generated promotion designed to conceal the operators' Russian origins.
The campaign reached limited audiences. The system it built is the bigger story.
AI did not invent the entire operation. It helped an elaborate false institution look active, credible, and ready to scale.
How the Operation Worked
OpenAI says the operators used ChatGPT to produce English and German social posts for X, LinkedIn, Facebook, Substack, and Telegram.
The prompts were often written in Russian. The operators explicitly asked the model to remove linguistic clues that could reveal where they came from. Because OpenAI does not provide model access from Russia, the accounts used VPNs.
The generated posts promoted articles from the International Burke Institute website. Some appeared under institute branded accounts. Others came from accounts posing as ordinary users.
The operation also produced replies to real Substack users, asking readers to follow the institute. A separate network of Telegram channels targeted audiences in Germany, the United States, France, Poland, and Türkiye.
This was not simply a flood of disposable AI content. It was a distribution layer around a carefully constructed identity.
The Most Disturbing Number Is 34 Out of 36
OpenAI reviewed a sample of 36 articles published on the institute's website between September 2025 and May 2026.
Thirty four were copied from elsewhere on the internet.
Some articles were years old. Others were credited to the wrong authors. One paper about the China Pakistan Economic Corridor appeared to come from Cambridge University Press but was attributed to a different academic.
Another article on migration policy was reportedly misattributed to an Australian professor of food chemistry.
The copied work gave the site a convincing surface. Real scholarship supplied the texture of legitimacy, while false names and institutional branding obscured its origin.
That combination is more dangerous than obvious spam because readers are trained to trust the signals it imitated.
A Ranking System Turned Propaganda Into Data
The institute also promoted a proprietary sovereignty index.
Its reports praised Russia while criticizing Western governments, the European Union, and countries that opposed the war in Ukraine. Political narratives were packaged as research and presented through the visual language of analysis.
This is a powerful tactic. Opinions feel more objective when they arrive as scores, charts, institutes, and expert commentary.
AI makes the promotional layer cheaper. Copied work makes the website look established. A custom index gives the operation something original to circulate.
Together, those elements can manufacture authority before a reader ever asks who is behind the institution.
The Campaign Was Small, but the Blueprint Is Scalable
OpenAI says most social posts received few views and the official institute accounts had limited audiences.
Its Telegram channels were more successful, generally attracting between 10,000 and 20,000 followers each.
Using the Brookings Breakout Scale, OpenAI placed the operation at the lower end of Category Three, meaning it appeared across multiple platforms and showed some signs of reaching authentic audiences.
That does not make it harmless.
The operation assembled durable infrastructure: a website, expert profiles, republished research, a proprietary index, branded channels, and a system for producing promotional content in multiple languages.
Once that machinery exists, volume can rise quickly.
What This Means for Every Reader
The old advice to check the source is no longer enough. A source can now be an entire fabricated institution.
Before trusting a polished research site, look for signals beyond design:
- Search exact sentences from articles to find copied material.
- Verify authors through university or employer pages.
- Check when the domain was registered and whether its history matches its claims.
- Examine whether experts have independent public records.
- Treat proprietary rankings cautiously when the method is vague.
- Look for unnatural language that may reveal translation or coordinated production.
The presence of real papers does not prove that the institution publishing them is real.
The Bigger Warning
Generative AI is often blamed for creating fake text.
This case shows a more sophisticated risk. AI can become one component inside a larger credibility machine built from stolen truth, professional design, invented expertise, and coordinated distribution.
OpenAI says the use of ChatGPT helped investigators expose the wider operation. That is encouraging, but the underlying economics remain unsettling.
It has never been cheaper to make a new institution appear busy, global, and authoritative.
The next influence campaign may not look like a bot network.
It may look like a think tank.
Read OpenAI's full investigation
The NEXAIUM Team
You follow the future. We decode it.
Quality Control
- Email and web pending approval
- No advertisement selected
- No forbidden long, medium, or double dash punctuation
- Fixed signature included
- Primary source: OpenAI, August 25, 2026