# U.S. and China set up AI risk dialogue, Bessent says

> The two governments plan to share testing practices and alert each other to serious AI incidents, while continuing to compete in advanced AI.

Oossa · 2026-11-11 · https://oossa.com/en/u-s-and-china-set-up-ai-risk-dialogue-bessent-says

The U.S. and China have established a dialogue on advanced AI risks, Treasury Secretary Scott Bessent said in prepared remarks to the National Committee on U.S.-China Relations in New York. The mechanism is intended to help the two governments manage risks from what Bessent called “super intelligence,” or SI, without ending their competition.

Bessent said the countries would exchange best practices for testing advanced AI models before release, keep human judgment in control, and alert each other to serious incidents involving the systems. He said the aim was to reduce shared risks and avoid misunderstandings between the two countries’ approaches.

## Cooperation has limits

Bessent did not commit the U.S. or China to slowing AI development. He said the countries should “harbor no illusions” about their differences, adding that their interests and approaches will not always align. The dialogue is about managing risks, not setting a shared ceiling on how advanced models can become.

The arrangement comes as Washington and Beijing compete for influence in AI. The U.S. has restricted exports of its most advanced chips to China and accused Chinese developers of copying U.S. models through what it calls industrial-scale distillation. The Trump administration has resisted broad AI restrictions, concerned they could weaken the U.S. position. President Donald Trump recently said he did not mind guardrails, but did not want regulation to put the U.S. “in a deep second position.”

## Leaders are due to meet next week

The dialogue follows an agreement during Chinese President Xi Jinping’s September visit to launch an AI discussion and a communications channel for defusing serious incidents. Bessent is expected to join Trump for a meeting with Xi on the sidelines of the Asia-Pacific Economic Cooperation Leaders’ Meeting, scheduled for November 18 and 19.

China has also pushed back on calls to slow frontier AI models. In September, its foreign ministry said fearmongering and confrontation would disrupt global AI governance. Chinese Ambassador Xie Feng said cooperation may not resolve every challenge, but that none could be resolved without it.

The two governments have not announced a binding agreement to limit AI development. Bessent described plans for sharing practices and incident alerts, but gave no further details about how the dialogue will operate.

## The facts

- Bessent said the U.S. and China had established a dialogue on risks from advanced AI.
- The planned exchanges include model-testing practices and alerts about serious AI incidents.
- The Asia-Pacific Economic Cooperation Leaders’ Meeting is scheduled for November 18–19, 2026.
- Bessent made no commitment to limits on AI development.

## Why it matters

For people who use AI tools, the planned exchange could help the U.S. and China respond to serious incidents involving advanced models. But the announcement does not set shared rules or slow development, so its practical effect remains unclear.

## Sources & references

1. [U.S., China should ‘harbor no illusions’ as they cooperate, compete on AI, Bessent says](https://www.cnbc.com/2026/11/11/trump-xi-summit-ai-nvidia-.html) – CNBC, 2026-11-11

Last updated: 2026-11-11