Pentagon Bars Anthropic From Military AI Work Over Safety Guardrails

The Pentagon designated Anthropic a supply chain risk in May 2026, effectively prohibiting federal contractors from deploying Claude in military contract work. The move reflects a clash between Anthropic's stricter safety constraints on military AI and the Pentagon's operational flexibility requirements. Seven other AI vendors—including OpenAI, Google, and Microsoft—secured access to classified networks. The exclusion signals that military priorities now outweigh the cautious AI safety approaches some companies advocate.