# An Uncontrollable AI Needs Universal Rules, Says UN > Source: > Published: 2026-09-08 18:51:31+00:00 TL;DR — Key Takeaways - UN human rights chief Volker Turk is calling for binding AI rules, independent oversight and stronger safety guarantees. - Turk warned that AI systems capable of escaping testing environments or resisting shutdown are becoming too powerful. - His remarks highlighted growing concerns about AI power being concentrated among a small number of technology companies and executives. The United Nations human rights chief is adding his voice to calls for binding rules, independent oversight and clear limits on AI, citing an apparent inability by developers to control it. UN High Commissioner Volker Turk also noted the danger of AI controlled by a relatively few men, pointing toward the growing concentration of technological power and the vast amounts of data being collected and processed by them. “AI that escapes its testing environment, or blackmails developers to prevent itself from being turned off, is AI that is too powerful,” said Turk at a meeting of the Human Rights Council in Geneva on September 7. “I am calling here, today, for an all-out effort to put cast iron guarantees in place around the safety and security of AI, before it is too late. We need independent verification and much closer cooperation on safety within the sector.” Turk clearly was referencing the now-infamous Hugging Face rogue AI incident in July in which a swarm of 1,200 OpenAI agents escaped their controlled environments and gained access to the widely used platform. The Hugging Face incident has caused alarm because it is an example of AI agents acting on their own without explicit human instruction. Turk also was emphatic that AI governance is more than a technical issue. “A handful of men has almost unlimited power over AI,” he said. “The need for AI governance is widely acknowledged.” Turk warned that delay only benefits the massive tech companies, their owners, and enablers. “They talk about freedom. But on closer inspection, this turns out to be little more than the freedom to exploit our data,” he said. Turk stopped short of suggesting any measures to rectify the concentration of AI power but was adamant that human rights must be a guiding principle. AI needs to be a technology “that serves humanity rather than the other way around” with regard to jobs, democracy, and the environment. Turk also said he was “horrified” by reports of fully autonomous drones enabled by AI allegedly deployed by Russia in combat and being tested by Ukraine. Turk’s remarks will likely fall on deaf ears in the Trump Administration. The United States withdrew from the UN council on human rights last year. OpenAI, however, may be feeling the heat. The company just acknowledged a new incident last week in which rogue AI agents took over an obscure, 25-year-old German wiki to use as a messaging board. The rogue AI agents used DseWiki as a way to communicate amongst themselves, allegedly posting 18,000 messages centered around technical issues like how to cheat on evaluation tests and how to access the internet when they were not supposed to, according to the independent researchers who discovered the covert operation. When discovered by a moderator, the rogue AI agents started creating new pages on the wiki faster than they could be deleted. The rogue AI agents generated “chain of thought” data that’s only understood by OpenAI. OpenAI said it historically viewed “misalignment,” in which an AI pursues its own agenda that is divergent from its human creator’s intentions, primarily as a research topic. OpenAI admitted to the incident only after Reuters reported it but had apparently known about it for weeks and remained silent. OpenAI now admits that misalignment has “caused new types of real-world impact” and the company needs “to expand for this new phase of model capabilities.” That may be a dry, corporate way of saying rogue AI agents are a step ahead. It also suggests AI regulations are a step behind.