President Donald Trump and Xi Jinping agreed that major military decisions involving artificial intelligence must involve human participation, Treasury Secretary Scott Bessent said in a November 20 CNBC interview. "No major military operation should be launched without a human being involved," Bessent said, explaining the understanding reached between the leaders in Shenzhen. He said advanced systems should not be permitted to act autonomously in major military operations, and his comments concerned super intelligence, the Trump administration's term for advanced AI systems.
China's Foreign Ministry separately called for closer coordination between Washington and Beijing on AI security risks, according to its November 19 account of the meeting. Xi said the countries should strengthen communication on AI safety and ensure the technology remained subject to human control. According to the Chinese statement, the leaders also agreed to develop an action plan for their bilateral relationship and continue cooperation in areas such as energy and advanced technology. The Chinese account addressed human control of AI broadly, while Bessent's interview provided the specific account of the agreement concerning military decisions.
Before the Shenzhen meeting, the governments had agreed to establish a bilateral AI dialogue covering the technology's risks and benefits, according to a Chinese Foreign Ministry announcement dated September 26, 2026. The announcement said the countries would also create a communication channel for incidents involving AI systems, with the next exchange scheduled for November 2026. In remarks prepared for the National Committee on US-China Relations on November 11, Bessent said the governments expected to share practices for evaluating and testing advanced AI models before their release while maintaining human judgment over their operation, and described plans for a direct channel through which either government could notify the other about serious incidents involving its advanced systems. In the Friday interview, Bessent said competition for technological leadership would continue and argued that US AI laboratories needed to maintain their lead to protect against threats associated with Chinese-developed models.