UN says AI safeguards can’t wait for certainty

The United Nations is telling governments not to wait for perfect certainty before acting on AI. In its first major assessment of the issue, a scientific panel says increasingly capable AI agents need to be reined in before their risks are fully understood.

That warning lands at a useful moment for the UN. Leaders are in New York for the General Assembly, and the US and China are also holding talks on AI. So the report is doing more than filing into a drawer; it is helping push AI squarely onto the diplomatic agenda.

The timing lines up with what UN secretary general António Guterres said last week. He urged governments to cooperate on AI threats and warned that the world cannot afford a race to the bottom on AI safety.

The report is also being framed as the UN’s first major assessment of OpenAI’s hack of Hugging Face earlier this year, which gives it extra weight beyond a generic policy paper. The message is simple: act before the system gets more capable, not after the damage is obvious.