As U.N. leaders prepared to discuss artificial intelligence in New York, the diplomatic gravity lay elsewhere: Xi Jinping was expected in Washington for talks with President Trump. The split-screen moment captured a hardening reality—AI governance is being shaped less by global institutions than by the rivalry between the United States and China.¹

The U.N. has tried to frame the technology as a shared danger requiring shared rules. Secretary-General António Guterres has placed AI alongside climate change and inequality among humanity’s existential risks, warning of disinformation, discrimination, privacy violations, cyberattacks and systems that could escape human control. “The world cannot afford a race to the bottom on A.I. safety,” he said.¹

On Monday, 22 countries adopted a declaration calling for stronger human oversight and control of AI systems. But the absence of its two most consequential players was conspicuous: neither Washington nor Beijing signed it.¹

Instead, the powers at the center of the AI race were pursuing a narrower channel. U.S. and Chinese officials discussed creating a mechanism to notify one another of AI-related national-security risks, to be called the U.S.-China A.I. Dialogue, according to Treasury Secretary Scott Bessent.¹ Yet the prospective cooperation sits beside deep disputes: Washington accuses Chinese firms of intellectual-property theft, while Beijing objects to U.S. limits on advanced chips.

Kai-Fu Lee of Brookings distilled the imbalance bluntly: “Frankly, the U.N. has been sidelined in the A.I. conversation.” In his view, AI has become “a two-player game,” akin to nuclear competition, in which countries without comparable capabilities struggle to shape the rules.¹

The U.N. may still have a role—not as referee over the superpowers, legal scholar Julian Ku argued, but as the institution representing countries that risk being shut out of AI’s benefits. That is a narrower mission, but perhaps the more realistic one.¹