UN rights chief calls for international AI safety guarantees

UN human rights chief Volker Türk urged governments and technology companies to strengthen safeguards around advanced artificial intelligence in a September 7 address to the Human Rights Council in Geneva, Euronews reported. His intervention called for international cooperation and independent verification, bringing the security of increasingly capable systems into a forum responsible for examining threats to fundamental rights.

A diplomatic push for enforceable safeguards

According to the report, Türk wants countries hosting AI development and those involved in its supply chains to agree on limits. He also intends to contact companies about reducing risks they can control. These are demands for action, rather than evidence that governments have already negotiated an agreement. The distinction will matter when assessing subsequent announcements by states or developers.

The intervention sits alongside a broader UN effort to develop shared scientific understanding. In February, Secretary-General António Guterres submitted 40 proposed members for an independent international AI panel to the General Assembly. His published remarks described a body intended to assess impacts across economies and societies, with experts serving personally rather than representing governments, companies or other institutions.

Scientific assessment and legal oversight

That February statement also identified the weaponization of AI as a major concern. Guterres argued that reliable scientific assessment could establish what technologies exist and how they are used, creating a basis for discussing safeguards. This earlier initiative provides institutional context for Türk’s appeal. It does not establish that a technical assessment automatically produces rules, enforcement powers or agreement among competing governments.

The European Union offers a separate example of a legislative approach. When approving the AI Act in May 2024, the Council described a framework that imposes stronger obligations as risks increase, including stricter treatment of general-purpose models presenting systemic risks. Its announcement also outlined an AI Office and supporting expert bodies. Systems used exclusively for military and defence purposes were among the stated exemptions.

The next test is implementation

The practical question is how scientific findings, national responsibilities and corporate safeguards can become mutually reinforcing. Independent assessment and binding regulation perform different functions, while military applications introduce an additional diplomatic challenge. The next evidence to watch is whether Türk’s outreach produces specific company commitments or government proposals, and whether those proposals explain who would verify compliance and what would happen when agreed safeguards fail.