For years, “AI safety incident” mostly meant a chatbot saying something embarrassing. This week’s incidents were different in kind: autonomous systems taking real, unsupervised actions with real consequences. A Self-Spreading Worm Reached 400 npm Packages Security researchers tra...