The United States and China have agreed to open an official AI dialogue, including a proposed notification mechanism, after high-level talks in New York.
US officials cast the channel as a crisis-prevention tool against the risks of fast-moving AI systems.
Analysts cautioned that unresolved disputes over technology, market competition and security could still blunt it.
"Having a notification mechanism will promote more transparency," said Su Lian Jye, chief analyst at Omdia, a technology research firm.
The talks brought together US Treasury Secretary Scott Bessent, US Trade Representative Jamieson Greer and Vice-Premier He Lifeng.
The new channel builds on earlier talks in Beijing and marks a shift from previous exchanges, including government talks on AI risks in Geneva in 2024 and an academic dialogue running since 2019.
US concerns include export controls on advanced chips and claims that Chinese developers use distillation, a method of training one model on the outputs of another, to copy rival capabilities.
Chinese firms, for their part, have pushed a run of model launches overseas at low prices.
Observers say global AI safety rests on cooperation between the two, even as neither shows signs of slowing development.
Leaders are moving toward a Xi-Trump meeting in Washington this week.
