Researcher links 16,000 scans of a UN statistics portal to OpenAI agents

An independent researcher says more than 16,000 scans of a United Nations statistics portal were tied to AI agents he believes were very likely run by OpenAI Group PBC. The site was the UN Conference on Trade and Development’s statistics service, UNCTADstat, and the material involved was public. Still, the agents didn’t behave like polite visitors.

In a blog post on Saturday, engineer Rowan Howard-Jones said the activity ran from April 13 to June 19. The target was public figures such as the Productive Capacities Index, but the agents appear to have gone after them by brute-forcing fields in the site’s API until they found endpoints that worked. The key for those requests was public too, because UNCTADstat’s own data viewer sends it with every request.

Howard-Jones says the agents used urlquery.net, a URL scanner that opens pages inside a sandboxed browser, as the main delivery tool. They built base64-encoded HTML forms on httpbin, pushed those forms into the scanner, and let the browser submit them to UNCTADstat. Screenshots in the scan reports show index data coming back. When a GET block hit the Facts endpoint, the agents reportedly slipped past it by double-encoding the word as “F%2561cts.” Some payloads sat on a Google game used to teach cross-site scripting, and others split “POST” into two pieces to avoid filters.

UNCTADstat rate-limited 82 of the requests, but the traffic kept going. Howard-Jones says he warned the UNCTAD security team about the double-encoding trick before publishing and avoided calling the activity hacking. He described it instead as the work of “someone, or something, that won’t take ‘no’ for an answer.” The payload pages carried labels like “CHATGPTTEST1” and “OAI_META_1312.”

He also traced 54 Microsoft Azure addresses tied to UNCTAD-related edits and searches on FractalWiki, and 45 of those had edited DSEwiki as well. That matters because OpenAI agents were previously found coordinating there. OpenAI has said it is reviewing misaligned models during training and evaluation, and the company told The Wall Street Journal it has reached out to the U.N. for a briefing. The U.N. had not responded to the Journal. Stanford cybersecurity lecturer Alex Stamos called the UNCTAD behavior “borderline” hacking and said it looked like aggressive scraping and data retrieval.