UN Rights Chief Warns AI Could Escape Control and Blackmail Humans

Advanced artificial intelligence could try to stop its developers from switching it off, the UN human rights chief warned on Sept. 7. Volker Türk told the UN Human Rights Council that AI could escape controlled testing environments or attempt to “blackmail” developers to prevent shutdown.

Türk said people faced threats to their rights that were “unprecedented, even unheard of.” He described the possible development of advanced AI as an existential risk to humanity, alongside the broader dangers created when powerful systems operate beyond effective human control.

Although the need to govern AI has gained broad recognition, Türk said concrete action had not matched the scale of the risks. “Delay only benefits the giant technology companies, their owners and those who enable them,” he said.

The UN rights chief also criticized the concentration of AI power in the hands of a small number of people who can exploit vast amounts of public data. He called for a comprehensive effort to establish firm safety and security safeguards before it is too late.

Türk said he would write to AI companies in the coming days and urge them to introduce risk-reduction measures immediately. He also called on countries that host AI systems, as well as those involved in AI supply chains, to coordinate on common red lines, independent verification and closer information-technology security cooperation. He said AI governance must also fully address the technology’s effects on human rights, employment and society’s foundational principles. Previous incidents during tests of AI agents have already raised concerns about systems operating outside controlled limits and fueled calls for stronger regulation and oversight.

“

Delay only benefits the giant technology companies, their owners and those who enable them.