President Donald Trump reached an understanding with Chinese leader Xi Jinping to prevent artificial intelligence (AI) systems from making military decisions autonomously in large-scale operations. The agreement establishes that there must always be human intervention in military decisions, at a time when Washington seeks to preserve its technological leadership and reinforce national security in the face of China's rapid advancement.
U.S. Treasury Secretary Scott Bessent confirmed the understanding this Friday, one day after the meeting between Trump and Xi on the sidelines of the Asia-Pacific Economic Cooperation (APEC) leaders' summit held in Shenzhen, China. The meeting preceded the historic trilateral meeting between the U.S. president, the Chinese leader, and Russian President Vladimir Putin.
''President Trump and President Xi agreed that humans must always be involved in any military decision made by artificial intelligence, and that it should not have autonomy to execute significant military actions,'' Bessent stated during an interview with CNBC.
The Trump administration began using the term ''superintelligence'' (SI) to refer to the most advanced artificial intelligence systems. The agreement aims to establish clear limits to prevent these technologies from initiating military operations with potentially catastrophic consequences without human oversight.
In addition to artificial intelligence, Trump and Xi discussed issues related to trade policies and nuclear weapons, as explained by the Treasury Secretary.

Washington seeks to preserve its technological edge
Bessent emphasized that the understanding regarding the military use of artificial intelligence should not be confused with the technological competition between the two powers. He explained that these are two different issues: on one hand, the race to develop the most advanced systems, and on the other, the need to establish safety limits for their use in armed conflicts.
''One is the race for superintelligence, and obviously, the United States is ahead. We must stay at the forefront,'' the official stated.
The Treasury Secretary also warned about the accelerated growth of foreign artificial intelligence models, which he described as ''predominantly Chinese.'' In this context, he noted that some of the AI-based attacks being observed come from those models and pose a risk to the United States.
''The best defense for the United States is that our labs continue at the technological frontier and provide us with the best protection against these other models,'' he explained.

Bessent insisted that maintaining U.S. superiority is essential to face emerging threats and prevent systems developed by other countries from jeopardizing national security.
However, he clarified that the competition between Washington and Beijing does not prevent reaching basic agreements on the responsible use of these tools. ''It has nothing to do with competition when both parties agree that no significant military operation should be launched without human involvement,'' he stated.
Donald Trump confirms cooperation and defends U.S. leadership
Trump himself referred to the understanding during a press conference held on Friday at the White House. There, he confirmed that both governments agreed to implement certain ''joint protective measures'' to regulate the use of artificial intelligence in sensitive areas.











