Anthropic's Safety Firestorm: Weapon Blocks, Researcher Exodus, and the UK's AI Reality Check

Anthropic published a 154-page threat intelligence report on Thursday revealing it blocked five cases of researchers attempting to use its Claude AI models to develop biological weapons, plus six cases involving conventional weapons including missiles, armed drones, bombs, and targeting systems. In northern Yemen, operators split work across conversations and obscured the ultimate purpose to circumvent safeguards, writing guidance and flight-control software for guided rockets...