Anthropic drops AI safety principles after Pentagon ban

Anthropic abandoned its core safety principles amid a Pentagon ban on Claude, its AI model, following a Trump administration declaration that the company posed a national security risk. The administration cited Anthropic's refusal to remove safeguards against domestic surveillance and autonomous weapons as a breach of supply chain trust. OpenAI secured a contract to supply the Pentagon instead. A week after the ban, U.S. military sources confirmed continued use of Claude for targeting decisions in Iran.