AI nearly had US troops attack a Chinese ship. Nobody could check its work.

A near-miss incident involving U.S. troops almost attacking a Chinese ship highlights the severe risks of insufficient oversight in AI deployment. The absence of independent checks and reliable records means potential blunders could go unchecked, impacting national security and beyond. This underscores the urgent need for a robust system to monitor and verify AI actions, akin to aviation's flight data recorders, to prevent future catastrophic errors.