On a Sunday in New York, two of the world's most consequential economic officials sat down for what Treasury Secretary Scott Bessent described as successful talks — and left with something few observers had anticipated: a serious conversation about artificial intelligence safety. Bessent, speaking to reporters after meeting Chinese Vice Premier He Lifeng, confirmed that the two sides had discussed establishing a shared mechanism to notify each other when AI incidents arise that could threaten national security. The proposal is modest in scope but significant in symbolism, representing the first known bilateral framework of its kind between the United States and China. The timing is not accidental.

President Donald Trump and President Xi Jinping are scheduled to hold a summit in Washington later this week, and both sides appear eager to demonstrate progress on multiple fronts simultaneously. The talks in New York also covered the so-called Board of Trade, a process designed to identify potential tariff reductions on goods. That conversation carries its own urgency: the existing tariff truce between the two countries is set to expire on 10 November, raising the pressure on negotiators to show results before a damaging trade escalation resumes. The AI dimension of the talks reflects a broader tension running through Washington's technology policy.

Trump has largely dismissed calls from researchers and industry figures to slow the development of AI, arguing that the United States cannot afford to fall behind China in what he has framed as a winner-takes-all race. "Whoever wins in AI, wins," he said earlier this month. Yet agreeing to share incident notifications with Beijing implicitly acknowledges that unilateral speed carries risks that even the most competitive posture cannot ignore. The notification mechanism, if formalised, would mark a rare point of structured cooperation between two powers that have otherwise treated AI dominance as a zero-sum contest.