Anthropic’s safety-first AI collides with the Pentagon as Claude expands into autonomous agents

As Anthropic releases its most autonomous agents yet, a mounting clash with the military reveals the impossible choice between global scaling and a “safety first” ethos