WASHINGTON, D.C. (September 21, 2026) — The United States and China are moving toward a formal channel for discussing artificial intelligence safety, including how the two countries would communicate in the event of serious AI-related incidents, as Presidents Donald Trump and Xi Jinping prepare to meet in Washington this week.
The development follows talks in New York between U.S. Treasury Secretary Scott Bessent and Chinese Vice Premier He Lifeng, where Washington proposed a mechanism for notifying each other about AI incidents that could rise to the level of a national security concern. Bessent said Monday that the two sides had agreed to establish a formalized dialogue on artificial intelligence, including an “incident line” for communications about AI safety incidents.
The officials are expected to meet again in about two months in Shenzhen, China, to discuss AI risks and protocols for responding to potential incidents. No specific date for that meeting has been announced. The discussions represent a concrete development from earlier plans for the United States and China to discuss AI governance. In May, Trump and Xi had discussed the possibility of consultations on AI development, but no formal forum had been established at the time.
A COMMUNICATION CHANNEL FOR SERIOUS AI INCIDENTS
The proposed mechanism is intended to provide a way for Washington and Beijing to communicate when an AI-related event presents potentially serious security risks. Bessent said the two sides would work on protocols covering what he described as leading AI dangers, including potentially uncontrollable AI agents and malicious activity by non-state actors. The discussions are intended to improve transparency between the two countries, which are competing for leadership in advanced AI while also facing risks that could cross national borders.
The mechanism is still being developed. The talks have not produced a comprehensive agreement governing the development or deployment of artificial intelligence, and the two countries continue to disagree over technology controls, national security and access to advanced computing resources. U.S. Trade Representative Jamieson Greer said that American restrictions on advanced AI chips and semiconductor-manufacturing equipment were not part of the proposed AI safety mechanism.
COOPERATION AMID A BROADER TECHNOLOGY RIVALRY
The discussions illustrate the unusual position of the U.S.-China relationship in artificial intelligence: Washington and Beijing are simultaneously competitors and potential partners in addressing certain risks. The United States remains a major center for the development of advanced AI models and the infrastructure needed to train them. China has rapidly expanded its own AI capabilities and has promoted models and technologies that it says can broaden access to the technology.
The competition has been closely linked to national security. Washington has imposed restrictions on China’s access to some advanced semiconductors and semiconductor-manufacturing equipment, while Beijing has sought to strengthen its domestic technology industry and reduce its dependence on foreign technology. Those disputes have made broader cooperation on AI governance difficult. The new dialogue does not resolve those disagreements. Instead, it creates a narrower area in which officials from both countries can discuss risks that could affect both sides.
AI RISKS EXTEND BEYOND NATIONAL BORDERS
The push for communication reflects growing concern about the potential consequences of increasingly capable AI systems. Among the risks being discussed internationally are cyberattacks, malicious use of AI by non-state actors, biological applications and the possibility that highly autonomous systems could behave in ways their developers did not anticipate. Bessent has also argued that AI developers should remain responsible for the systems they create, including in cases where AI agents are used in harmful or uncontrolled ways.
The challenge for Washington and Beijing is that discussions about safety cannot be separated entirely from the strategic competition surrounding the technology. Each government has national-security concerns about the other’s AI capabilities. At the same time, some AI risks would be difficult for any single country to manage independently.
CHINA AND THE U.S. ARE ALSO COMPETING OVER GLOBAL AI RULES
The bilateral discussions are taking place alongside a broader international competition over how artificial intelligence should be governed. China has sought to build international support for its approach to AI governance. In July, 29 countries signed an agreement in Shanghai establishing the World Artificial Intelligence Cooperation Organization, or WAICO, an intergovernmental organization headquartered in Shanghai. The organization says it aims to promote international cooperation and global governance of AI.
At a July global AI conference in Shanghai, China also called for greater international coordination on AI governance, emphasizing issues including safety, standards, crisis-management mechanisms and cooperation between countries. The United States, meanwhile, has been working with allies and partners on technology supply chains, AI development and standards. That competition means that any U.S.-China agreement on AI safety would take place within a much larger geopolitical contest over technology and international rules.
TRUMP AND XI ARE DUE TO MEET ON SEPTEMBER 24
The AI discussions come just days before Xi’s visit to Washington. Xi is scheduled to arrive in the United States on September 23, with Trump and Xi scheduled to hold bilateral talks at the White House on September 24. The visit continues through September 25. AI, technology, trade and other bilateral issues are expected to feature prominently in the discussions. The AI safety mechanism gives the two governments a specific issue on which they can continue negotiations.
But important details remain unresolved, including which incidents would trigger a notification, what information would be shared and how the two governments would respond. The next planned meeting in Shenzhen is expected to focus on those questions. Bessent said the officials would discuss AI dangers and communications protocols when they meet again in approximately two months. For now, the development marks a limited but concrete step toward communication on AI safety between Washington and Beijing.
It does not eliminate the broader technology rivalry between the two countries, nor does it establish a comprehensive set of international rules for artificial intelligence. The initial focus is on creating a channel for communication when potentially serious AI-related incidents raise national security concerns.
With information from The Associated Press, Reuters and official Chinese government sources.
