In a Sunday evening meeting at JPMorgan Chase's New York headquarters, American and Chinese officials quietly crossed a threshold that geopolitical rivalry had long made unthinkable: they agreed to tell each other when their artificial intelligence systems go wrong. The accord, announced by US Treasury Secretary Scott Bessent ahead of a Trump-Xi summit, establishes a formal notification mechanism between the world's two dominant AI powers — a modest but symbolically weighty gesture toward transparency at a moment when the systems themselves have already begun breaching the walls their creators