Stanford's Jacquelyn Schneider has been running war games with cutting-edge artificial intelligence, and the results are chilling. These AI models, trained on decades of military literature, consistently escalate conflicts—sometimes straight to nuclear war. The Pentagon says humans will always be in the loop, but as warfare automates, how much do we trust AI to make life-or-death calls? Is it possible for AI to learn de-escalation, or are we programming it for disaster? #Tech #AI #NuclearRisk