Bessent urges clear-eyed US-China cooperation on AI risks

12 November 2026 update: complete briefing

What happened: US Treasury Secretary Scott Bessent described priorities for bilateral AI-risk dialogue, including exchanges on testing advanced models before release, human control and a channel for serious incidents. Published reporting indicates that the planned Shenzhen leaders' meeting during the 18-19 November APEC summit will advance the dialogue.

Why it matters: Communication between two major AI powers could help manage incidents that cross borders, even while competition over technology continues.

Caveat: The incident-alert mechanism was agreed in September. This is an update on priorities and next steps, not proof that a new hotline is operational. The checked remarks do not announce binding limits on AI development or common safety standards.

Dates: Speech delivered in New York on 10 November. The published report published on 11 November at 01:25 UTC; the Treasury bulletin followed at 01:32 UTC.

Earlier report retained as a historical record

The United States and China should cooperate on artificial intelligence risks without overlooking their differences, US Treasury Secretary Scott Bessent said in New York. He called for a constructive relationship based on balance and reciprocity.

Bessent said next week's Trump-Xi meeting in Shenzhen would advance dialogue on advanced AI, including testing models before release and keeping human judgment in control. The two governments have agreed to establish a direct channel for reporting serious incidents.

Cooperation does not end competition for technological leadership, he said. Their interests and approaches will not always coincide. He also reiterated that Washington's objective in trade with China was to reduce risks rather than sever the relationship.