Abstract:
U.S. Treasury Secretary Scott Bessent said that during their meeting in Shenzhen, U.S. President Trump and Chinese President Xi Jinping reached an important consensus on the application of artificial intelligence in the military field,
Both sides agreed that military decision-making involving artificial intelligence must always retain human participation, and artificial intelligence systems should not be allowed to make completely autonomous decisions in major military operations.

In an interview with the American media after concluding his trip to Asia, Bessant said that this consensus is one of the important results achieved by Trump's trip to Asia. Although China and the United States are competing fiercely in the field of artificial intelligence, both sides still believe that necessary security mechanisms should be established in military applications to ensure that key decisions are always in human hands.
According to Bessant, the two countries agreed that no major military operation should be launched without human participation and supervision. He emphasized that this principle is not inconsistent with the competition between the two sides in the field of artificial intelligence technology, because no matter how the competition intensifies, humans must retain final control over major decisions involving war and security.
When talking about artificial intelligence competition, Bessant said that it is still very important for U.S. artificial intelligence research institutions to maintain their technological leadership. He said that the United States needs to continue to innovate to deal with what he describes as potentially risky foreign artificial intelligence systems, and believes that technological advantages are still an important factor in ensuring national security.
In the interview, Bessant also mentioned the recent controversy surrounding former Nvidia CEO Jensen Huang. He said that the relevant accusations involving Huang Renxun were not mentioned in the talks between Trump and Xi Jinping, nor were they part of the bilateral discussions between the two parties.
On the issue of chip exports, Bessant said that Nvidia H200 is still one of the most advanced artificial intelligence chips that China can purchase under license conditions. He said that the U.S. government has taken a tough stance on the export control of advanced artificial intelligence chips, and Chinese companies have also been working hard to build their own chip industry system in recent years.
Bessent believes that China still has a clear gap with the United States in the field of high-end artificial intelligence chips, and said that the current demand for H200 chips from the Chinese market is not strong, partly because Chinese companies are promoting the construction of independent chip ecosystems.
Bessent downplayed the different voices within the U.S. government on the chip export restriction policy. He said that although the heads of the Treasury and Commerce Departments had different views on some issues before, all departments of the Trump administration will ultimately implement the policy directions set by the president.
Earlier this week, the U.S. Department of Commerce further tightened restrictions on China’s acquisition of advanced artificial intelligence chips. This includes not only the export of physical chips, but also indirect channels for acquiring advanced computing power through leasing computing power in overseas data centers. Bessent said that relevant restrictions are being gradually implemented, and remote access to high-end artificial intelligence computing resources will also be more restricted.
Analysts believe that as the competition in artificial intelligence between China and the United States continues to heat up, the two sides have publicly announced for the first time that they have reached a principled consensus on the governance of military artificial intelligence. This shows that both countries are aware of the security risks that may be brought about by the rapid development of artificial intelligence technology, and are trying to establish a minimum security boundary for artificial intelligence applications in the military field outside of strategic competition.
Comments