Trump and Xi Agree Human Oversight Must Remain in Military AI Decisions, Bessent Says
The reported understanding marks a potential area of cooperation as Washington and Beijing compete over artificial intelligence and debate safeguards for autonomous weapons.

WASHINGTON — U.S. President Donald Trump and Chinese President Xi Jinping have agreed that humans should retain control over military decisions involving artificial intelligence, including major military actions, according to U.S. Treasury Secretary Scott Bessent, highlighting a potential area of cooperation between the two countries amid their broader technological competition.
Bessent described the understanding as one of the important outcomes of Trump's meeting with Xi in Shenzhen, China, on Thursday, during the Asia-Pacific Economic Cooperation (APEC) Economic Leaders' Meeting.
Speaking to CNBC after returning from Trump's Asia trip, Bessent said the two leaders had agreed that artificial intelligence should not independently make major military decisions.
According to a Reuters report carried by Investing.com, the agreement would require human involvement in military decisions using AI technologies and prevent autonomous systems from initiating major military actions.
The reported understanding comes as both Washington and Beijing accelerate the development of artificial intelligence for military, intelligence and national security applications.
AI technologies are increasingly being integrated into surveillance systems, battlefield information processing, autonomous vehicles, logistics planning and decision-support tools. The growing capabilities of these systems have raised concerns about accountability, accidental escalation and the possibility that automated systems could make decisions faster than human operators can adequately assess them.
The reported agreement suggests that the two governments recognize at least some shared concerns about the risks associated with military AI, despite their differences over technology policy, semiconductor exports and strategic competition.
However, the details of the understanding remain limited. Neither government has publicly released a comprehensive agreement specifying which military AI systems would be covered, how human oversight would be defined or what verification mechanisms might be established.
Bessent also emphasized that the United States continues to compete with China in artificial intelligence. He said maintaining American technological leadership remains a priority for Washington, while describing the broader relationship as one in which the two countries seek to reduce risks without abandoning competition.
The Shenzhen meeting was the third face-to-face meeting between Trump and Xi in 2026. Their discussions covered trade, investment, security and bilateral relations, according to official statements.
Xi said the two countries should strengthen communication and coordination on AI safety risks and ensure that artificial intelligence remains under human control, according to China's official meeting readout.
The reported military AI understanding comes against the backdrop of growing international debate over autonomous weapons systems and the rules governing their use.
Governments, researchers and international organizations have increasingly called for safeguards to ensure that military AI applications remain accountable and comply with international humanitarian law.
For China and the United States, the challenge will be translating the reported political understanding into practical guidelines that military institutions can implement.
The agreement could provide a foundation for future discussions on AI risk management, crisis communication and military technology safeguards. Nevertheless, without publicly available technical details or formal implementation arrangements, its immediate significance remains primarily diplomatic.
The development highlights a potentially important area of cooperation between the world's two largest economies as they seek to manage emerging security risks while competing for leadership in advanced artificial intelligence.

