UN says AI safeguards can't wait for certainty

A UN scientific panel urged governments to rein in increasingly capable AI agents before their risks are fully understood, in the body's first thematic brief. The report lands as the US and China hold AI talks and leaders gather in New York for the UN General Assembly.

• The brief is the first from the UN's Independent International Scientific Panel on AI, set up last year
• The panel invokes the precautionary principle from the 1992 Rio Declaration
• It follows documented AI agent incidents at OpenAI, Anthropic, Google and Meta
• UN chief António Guterres warned against a 'race to the bottom on AI safety'