Bill Gates says the United States and China will ultimately have to cooperate on governing artificial intelligence, even though both countries are competing to build it.
His proposal begins with new domestic and international institutions capable of addressing AI across multiple sectors. Gates points to risks such as cybercrime, disinformation, deepfakes, bioweapon assistance and autonomous weapons as problems that can cross borders.
Gates uses nuclear inspection and aviation regulation as reference points, but those analogies do not make AI equivalent to either technology. His point is narrower: some globally consequential systems require coordination beyond ordinary national regulation.
The difficulty is cooperation without trust. Governments may share an interest in preventing catastrophic outcomes while withholding information, protecting strategic advantages and resisting rules that could slow their own developers.
Gates’s essay proposes an institutional direction, not a complete treaty, verification system or enforcement architecture. The unresolved question is whether rivals can agree on a small enough set of shared risks to begin governing a technology they are still racing to control.