Chinese President Xi Jinping and US President Donald Trump reached a shared understanding at the Asia-Pacific Economic Cooperation summit in Shenzhen that human judgment must remain central to any major military decision involving artificial intelligence, US Treasury Secretary Scott Bessent said on Friday.
In an interview with CNBC, Bessent described the agreement as a breakthrough from the leaders' meeting on the sidelines of the APEC gathering. Both sides, he said, agreed that humans should always be involved in military decisions made with superintelligence and that AI should not be allowed to act autonomously on major military actions. Trump has used the term superintelligence, or SI, to describe advanced AI.
A guardrail, not a concession
Bessent was quick to separate the safety agreement from the broader technology competition between Washington and Beijing. He said the guardrail had nothing to do with competition and was focused instead on managing risk.
At the same time, Bessent stressed that American AI developers must keep their technological edge, calling it the strongest defence against what he described as dangerous foreign models, noting that these were predominantly Chinese. The comments underline that the consensus on AI safety does not mean any easing of US technology restrictions.
From bilateral to trilateral
The Xi-Trump meeting was followed by an unprecedented trilateral summit with Russian President Vladimir Putin. The three leaders later issued a joint peace declaration covering cooperation on global energy markets, nuclear non-proliferation and global governance.






