Anthropic agreed in December to let the Pentagon use its AI for missile and cyber defense, but the company’s restrictions on broader military applications have triggered a showdown with Defense Secretary Pete Hegseth. According to a senior Pentagon official, Hegseth issued an ultimatum to CEO Dario Amodei to permit “any lawful use” of the AI systems by Friday or face compelled cooperation under the Defense Production Act and a “supply chain risk” designation that could cut the firm off from defense work. Anthropic, which emphasizes AI safety and bars applications like mass surveillance and lethal autonomous weapons, says its proposed contract language already enables missile-defense use. The standoff comes as the Pentagon pursues an “AI-first warfighting force” and expands classified-network access to commercial AI, with xAI agreeing this week to the Defense Department’s terms. Anthropic is among four firms awarded up to $200 million to prototype frontier AI for national security, heightening the stakes for how Washington balances safety guardrails with military flexibility.
Related article:
