The world's two leading artificial intelligence powers, the United States and China, are struggling to cooperate on managing dangers from the technology described as potentially existential. Existential, in plain terms, means risks severe enough to threaten the foundations of human civilization. Technological rivalry between Washington and Beijing is what makes agreement hard.
Technological rivalry, as a concept, goes beyond economic competition. It describes a situation where two countries are each trying to become the dominant power in a field that carries consequences well beyond commerce, including consequences for military strength and geopolitical influence. When AI is that field, the race is one that neither side can afford to lose, and that logic shapes every interaction the two governments have.
The competition creates a direct conflict with safety cooperation. When two countries race to lead in AI development, the information required to manage safety risks is often the same information that reveals how advanced each side's systems are. Sharing it means giving a competitor insight into your technical capabilities. For rivals in a high-stakes race, that trade-off is genuinely difficult to accept.
Governing dangerous technologies generally requires governments to be transparent with each other about what they know and what they fear. That transparency usually means acknowledging what each side can do and accepting that doing so reveals vulnerabilities to the other. AI presents that same basic requirement, but with a harder version of the same difficulty. Technical knowledge in AI development translates directly into economic and military advantage, which means the incentive to withhold is stronger.
Existential danger from AI, as a category, is not a claim about any single product or company. It refers to outcomes severe enough that they could reshape or end human civilization as it exists. The scale of that concern is also what makes international coordination necessary: these risks, if they develop, would not stop at national borders.
The tension is structural. Two governments that view each other as technological rivals need to find enough common ground to set safety rules for a technology neither fully controls. Agreement would require each side to accept some degree of transparency about its own capabilities and some degree of trust in the other's intentions. Washington and Beijing have not yet found a way to make that trade.