Washington and Beijing have agreed to launch AI discussions ahead of a crucial summit between Presidents Donald Trump and Xi Jinping, scheduled for Thursday, Sept. 24, in Washington. Dire warnings about the threat of unchecked AI have injected a fierce sense of urgency into the engagement.

Until now, attempts to establish global guidelines for AI safety have repeatedly stalled, hampered by intense competition and distrust between the world's two superpowers. Widespread doubt about the technology's existential risks has also delayed progress: Trump has openly rejected such warnings, a position that experts say aligns with China's broader strategic approach. That the world's two leading AI powers are opening a channel at all gives the engagement significance beyond its modest immediate agenda — communication, not treaty-making, is the starting point.

In one of his more recent posts on X, Trump wrote, "The people that say AI is going to destroy the World, and that Data Centers are bad for your neighborhood, are the same people that said, just a short time ago, that the World would be extinguished by 'Climate Change. That HOAX never worked out for them, and now they're on to the next one." (via FT on X)

Despite deep-seated friction, the necessity for U.S.-China coordination has taken on new prominence ahead of the leaders' Sept. 24 summit in the U.S. capital. The shift follows unprecedented warnings from the American tech industry demanding a slower, more deliberate approach to AI advancement.

Bessent says U.S. proposed a notification system for AI failures or incidents

The U.S. administration has already begun laying the groundwork for the discussions. Trump confirmed that AI is on next week's packed schedule, and Treasury Secretary Scott Bessent told reporters that the United States is willing to discuss mutual risks with Beijing. Bessent also revealed that Washington has floated a formal notification protocol for AI failures or incidents.

"We think that just like any cross-border activity that is moving from opaque to more transparency between the number one and number two AI powers in the world is very important," Bessent said.

The proposed mechanism would focus on some of the most serious risks associated with advanced AI, such as cyberattacks, system failures, and threats to critical infrastructure. The immediate aim, rather than a global AI treaty, is establish a way for the two countries to notify each other when an AI-related incident could threaten both nations' national security. Such a channel would allow Washington and Beijing to share information in the event of an emergency, even as the two governments continue to compete over advanced chips, AI models, and computing infrastructure.

Some U.S. AI company executives, including Anthropic chief Dario Amodei and OpenAI's Sam Altman, have called for a slowdown in AI development given the rising number of risks. Amodei has particularly advocated steps the industry could take in "pacing the frontier," and he has also backed Trump's vision that the U.S. needs to maintain its competitive edge. One former Anthropic employee, by contrast, offered a more alarming outlook, asserting that he worries about humanity's existence in the coming years.

Some U.S. lawmakers have so far insisted on the need to establish inspections and safety standards with China, while others continue to encourage Trump to take steps to slow China's AI development, including by limiting its access to sophisticated AI technologies. The split underscores the tension running through the engagement: pursuing safety cooperation with Beijing even as the race for AI advantage continues on parallel tracks.

Analysts see the broader rivalry as the biggest obstacle

On China's side, some have also cautioned about the technology's possible dangers. China's top intelligence official, for instance, has contended that artificial intelligence directly threatens the Chinese Communist Party's political authority. Chinese Foreign Ministry spokesperson Guo Jiakun, meanwhile, has emphasized that international stakeholders should work together to advance an open, inclusive AI framework dedicated to a global public good.

Alison Szalwinski, senior vice president at strategic and capital advisory firm The Asia Group, argued that while the U.S. and China now broadly agree on the dangers of AI cyber capabilities, the biggest obstacle is whether AI safety can be decoupled from overall U.S.-China hostility.

"At this point, the immediate goal is less about negotiating a major AI agreement — we're just not there yet, and more about establishing a durable channel for discussing frontier AI risk," Szalwinski said.

Kyle Chan of the John L. Thornton China Center at the Brookings Institution also expressed doubts about the prospects for cooperation on AI safety. "I don't think that Trump is feeling any pressure about the existential AI risk issue. He calls it a hoax multiple times. It reminds me very much of how he treats climate change," he said. Chan added that Trump is focused primarily on promoting AI demand and the stock market's rise.

The immediate markers for the engagement are concrete: whether the notification protocol advances from floated proposal to working channel, and what, if anything, the two leaders say about AI when they meet on Sept. 24.