In the shadow of an impending summit between two of the world's most consequential leaders, American and Chinese officials have begun sketching the outline of a shared language for artificial intelligence risk — a quiet but significant gesture toward cooperation amid fierce rivalry. Treasury Secretary Scott Bessent and Chinese Vice Premier He Lifeng met in New York to discuss a mutual notification system for AI incidents with national security implications, borrowing a framework long used in arms control and financial diplomacy. The proposal does not slow the race; it simply asks that each run