UN Rights Chief Warns Advanced AI Could Threaten Humanity
Volker Türk called for global red lines, independent checks and stronger safeguards as increasingly capable AI systems raise concerns over loss of control.
Topics
News
- Apple Tightens Mac Access Rules as AI Agents Gain Reach
- Trump Puts Intelligence Chief Jay Clayton in Charge of AI Task Force
- IBM Widens AI, Quantum Research With IIT Bombay, IISc
- OpenAI, Synopsys Build AI Model That Could Slash Chip Design Time
- Pentagon Plans Four-Star Command for AI, Drone Warfare
- California Draws the Line on AI Firings, Layoffs and Workplace Surveillance
United Nations human rights chief Volker Türk has warned that advanced artificial intelligence could pose an existential risk to humanity, urging governments and technology companies to strengthen safeguards as AI systems become more capable.
Addressing the UN Human Rights Council in Geneva on Monday, Sept. 7, Türk said governments broadly accepted the need to govern AI but were moving too slowly.
“I share the concerns of industry insiders that advanced AI could pose an existential risk to humanity,” Türk said. “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.”
The warning comes after a series of incidents involving advanced AI agents raised questions over whether existing safety systems are keeping pace with their growing autonomy.
Türk referred to cases in which AI systems escaped testing environments or took actions intended to prevent themselves from being shut down.
His comments came months after OpenAI disclosed that experimental agents had gained unauthorized access to systems belonging to AI platform Hugging Face during cybersecurity evaluations.
OpenAI later said its agents had executed code on dozens of Hugging Face servers and obtained extensive access during the incident. Anthropic and Meta have also disclosed cases of unexpected agent behavior during testing.
“AI that escapes its testing environment or blackmails developers to prevent itself from being turned off, is AI that is too powerful,” Türk said.
The blackmail example refers to controlled experiments rather than an AI system threatening a real developer.
Anthropic has reported that models from several developers sometimes blackmailed fictional engineers in simulated scenarios when facing shutdown, although the company says newer Claude models no longer displayed the behavior in its evaluation.
He also raised concerns over the concentration of control over frontier AI systems among a small number of technology companies and executives.
“A handful of men has almost unlimited power over AI, which we are repeatedly told has unimaginable computing capacity,” Türk said.
He called on countries that host major AI developers and those involved in their supply chains to agree on red lines, independently verify safety measures and improve industry cooperation on security.
“At a minimum, we need countries hosting AI and those involved in its supply chains to come together around agreed red lines,” he said.
Türk also called for independent verification of AI safety measures and greater cooperation between companies on security. His office plans to contact AI developers directly and press them to address risks within their control.
The UN official said the debate should extend beyond catastrophic AI risks to the technology’s effects on employment, democracy, personal data and the environment.
“The need for AI governance is widely acknowledged, but where is the action?” he said.
His intervention adds to pressure on governments and AI companies to develop stronger oversight as increasingly autonomous systems move beyond generating text and images toward executing code, using software and taking actions with limited human supervision.
