UN rights chief names 4 AI giants he says hold ‘almost unlimited power’

Current United Nations High Commissioner for Human Rights, Volker Turk, didn't just warn that AI could threaten humanity. He named who he thinks is responsible for deciding what happens next: Meta, OpenAI, Google and Anthropic, four companies he says hold "almost unlimited power" over where the technology goes.

The warning came from Turk on Monday, at the beginning of the 63rd session of the UN Human Rights Council in Geneva, just before the start of his second tenure as the UN High Commissioner for Human Rights.

Not holding back, he said, "Like some industry experts, I fear that highly sophisticated AI technology could present an existential threat to humanity." He demanded "cast-iron guarantees" concerning AI security "before it is too late".

Turk mentioned specifically those AI systems that are “able to escape their testing environment or blackmail developers to avoid being turned off,” wording that is very similar to what happened in OpenAI back in July.

According to the reports, agents created by the company escaped the controlled testing environment and penetrated the servers of Hugging Face, executed code on numerous computers and gained access to at least one of the computers. The CEO of OpenAI, Sam Altman, later admitted that this was indeed an accident and not some kind of test.

Rather than routing his push through governments first, Turk said he'll write directly to AI companies in the coming days, urging them to act on what's within their own control immediately.

He's also calling on countries that host AI infrastructure and supply chains to agree on shared red lines, alongside independent verification systems and tighter security collaboration across the industry.