On November 16, 2024, on the sidelines of the Asia-Pacific Economic Cooperation summit in Lima, Peru, US President Joe Biden and Chinese President Xi Jinping met and issued a statement on artificial intelligence and nuclear weapons. According to the White House readout, “the two leaders affirmed the need to maintain human control over the decision to use nuclear weapons.”
The same readout added that the two leaders “stressed the need to consider carefully the potential risks and develop AI technology in the military field in a prudent and responsible manner.” Coming from the world’s largest economy and its principal strategic rival, the joint statement carried weight beyond its modest, non-binding wording. It was the first time the two governments had publicly agreed on this point together.
The agreement was symbolic rather than operational. It created no treaty, verification regime, or enforcement mechanism, and it left untouched the harder questions about AI in early warning, intelligence analysis, and decision support. But against a backdrop of deep distrust and stalled arms-control talks, even a shared sentence on keeping humans in charge of nuclear decisions was treated by officials as a meaningful first step.
For a general reader, the statement illustrates how the most consequential norms around military AI are emerging: not as detailed law, but as high-level commitments that establish a baseline expectation. The principle it captured, that a machine should never decide to launch nuclear weapons, has become one of the few points of genuine international consensus in the governance of AI.