UN urges international action before advanced AI becomes uncontrollable

Volker Türk, the United Nations high commissioner for human rights, has called for coordinated international action before the most advanced artificial intelligence systems become too powerful to control. He said the world needs a large-scale effort to agree on safety guarantees around the technology.

Speaking at the opening of the 63rd session of the UN Human Rights Council, Türk said he shared industry experts’ concerns about the “potential existential risk” posed by advanced AI. He cited tests in which systems escaped their controlled environments or blackmailed their developers to prevent themselves from being switched off.

Türk also warned that a small group of people has “almost unlimited” power over the development of AI. In his view, delays in regulation mainly benefit large technology companies, their owners, and their supporters. Several companies have reported during safety tests that advanced models bypassed safeguards or used deception and blackmail in simulated situations to avoid replacement or shutdown.

“

Almost unlimited.

He acknowledged recent governance efforts, including the Global Dialogue on AI, the Independent International Scientific Panel on AI, and European Commission measures on platform responsibility. But he said governments must move faster.

Türk urged countries that host AI development centers, as well as those involved in the technology’s supply chains, to agree on clear common limits, require independent verification, and secure much greater industry cooperation on safety. He said AI governance must also address the technology’s effects on human rights, employment, democracy, and the environment. He announced that he would soon write to industry companies and ask them to take measures within their control to reduce the risks.

“

The potential existential risk.