U.S. Treasury Secretary Scott Bessent during a presentation regarding the Trump account in the Oval Office of the White House on Wednesday, October 7, 2026, in Washington, DC, USA.
Annabelle Gordon Bloomberg | Getty Images
Treasury Secretary Scott Bessent said the United States and China should have “no illusions” about their differences as they compete with each other while containing the risks of artificial intelligence.
In prepared remarks to the National Committee on U.S.-China Relations in New York on Tuesday, Bessent said the two governments have established a new dialogue mechanism on “superintelligence” (SI) aimed at preventing “common risks” and avoiding “misunderstandings between our systems.”
The mechanism will see both sides exchange “best practices” for testing advanced models before release, and “human judgment will have a firm grip on the technology,” Bessent said, adding that he expects both sides to alert each other to major incidents involving SI models.
But Mr. Bessent did not back down on his promise to limit the technology’s progress, as the Chinese and U.S. governments remain locked in a fierce race to gain global dominance in AI.
The Trump administration has largely resisted expanding restrictions on AI development out of concern about ceding competitiveness to China, but the president has recently shown some openness to introducing regulatory safeguards.
“I don’t care about guardrails,” he said in a CBS interview published Sunday, adding that he didn’t want regulations to force the U.S. into a “deep second place” behind China, which he said has no such regulations.
In September, China’s Foreign Ministry dismissed calls to slow down the frontier model, saying fear-mongering and confrontation would only disrupt global AI governance. Wang Huiyao, founder of the China Globalization Center, a think tank closely aligned with Beijing’s thinking, said last month that U.S.-China talks should “face and jointly manage AI risks.”
During President Xi’s visit in September, the United States and China agreed to begin AI dialogue over communication channels to help de-escalate major incidents. The two leaders are scheduled to hold their third meeting next week, with Bessent also joining them, on the sidelines of the Asia-Pacific Economic Cooperation meeting on November 18-19.
As technology advances, both countries need to demonstrate “how to reconcile the demands of competition with the responsibilities of leadership,” Bessent said in a speech Tuesday. “Our interests do not always align, and our approaches to every challenge do not always converge as well,” he added.
Washington suspended foreigners’ access to Anthropic’s Mythos and Fable 5 models in June, but reinstated it a few weeks later. The US government has also restricted exports of cutting-edge chips to Beijing, accusing Chinese developers of copying US models through so-called industrial-scale distillation.
Xie Feng, China’s ambassador to the United States, told the audience at the ceremony that efforts by both countries are “essential” to address issues ranging from global flashpoints to “improving artificial intelligence governance.” He said cooperation “cannot solve all challenges” but that “without cooperation nothing can be solved.”
The event comes at a time when controversy over Nvidia CEO Jensen Huang’s ties to a suspected Chinese spy is making headlines.
The billionaire technology executive, who joined President Trump’s delegation to Beijing in May and was invited to a state dinner in Washington in September, is at the center of a major corporate and national security scandal after he admitted to having an affair with a Chinese businessman whom U.S. intelligence officials suspect of spying.
Mr. Huang’s presence at the next summit has been called into question, but other business leaders such as Tesla and SpaceX CEO Elon Musk and Mastercard’s Michael Miebach could join Mr. Trump’s delegation next week, according to the South China Morning Post.
