Scott Bessent has predicted that US-China AI cooperation will have limits. The US Treasury Secretary said Washington and Beijing should remain clear about their differences while working together to manage risks from advanced artificial intelligence. In prepared remarks to the National Committee on US-China Relations in New York this week, he outlined a dialogue mechanism intended to prevent misunderstandings between the two countries.?“Our interests will not always coincide, and our approach to every challenge likewise will not always converge,” Bessent said, CNBC reported. His remarks placed cooperation on AI safety alongside continued competition for technological leadership. The dialogue includes exchanges on model testing and communication about serious incidents, without a commitment to slow AI development.?

What will the US-China AI dialogue cover

?Bessent said the two governments had established a dialogue on “super intelligence,” or SI, to address shared risks and avoid misunderstandings between their systems.?The mechanism would allow both sides to exchange practices for testing advanced models before release. He also emphasised “keeping human judgement firmly in command of the technology” and said he expected the countries to alert each other to serious incidents involving those models.?During the committee’s gala, Bessent argued that, as the technology develops, both countries would need to “reconcile the demands of competition with the responsibilities of leadership.” ?

Cooperation does not mean slowing the AI race

?Bessent did not commit to limiting advances in AI. Washington and Beijing remain competitors in developing the technology, even as they establish channels to discuss its risks.?The US government has resisted broader development restrictions, concerned they could weaken the US position against China. However, US President Donald Trump indicated some willingness to consider safeguards in a CBS interview released last week.?“I don’t mind guardrails,” Trump said, while adding that he did not want regulation to leave the US in “a deep second position” behind China. He said China had no such restrictions.?Meanwhile, in September, China’s foreign ministry dismissed calls to slow frontier models, arguing that fearmongering and confrontation would disrupt global AI governance.?

Talks come ahead of another expected Trump-Xi meeting

?The dialogue follows an agreement during Chinese President Xi Jinping’s September visit to establish AI talks and a communications channel for serious incidents.?Trump and Xi are expected to meet again on the sidelines of the Asia-Pacific Economic Cooperation Leaders’ Meeting on November 18–19, with Bessent joining.?At the New York event, China’s ambassador to the US, Xie Feng, also argued for cooperation on AI governance. Cooperation “may not resolve every challenge,” he said, but “none can be resolved without it.”