Pentagon vs Anthropic over Claude limits for autonomous arms

According to Associated Press, Pentagon CTO Emil Michael said he clashed with Anthropic over the use of its AI in fully autonomous weapons, a dispute that followed a debate about how AI could support a Golden Dome missile defence programme designed to place U.S. weapons in space. Michael described Anthropic’s ethical restrictions on Claude as an irrational obstacle as the military moves toward greater autonomy for swarms of armed drones, underwater vehicles and other machines to compete with rivals.

He said negotiations with Anthropic lasted three months and involved demanding that terms of service be rational relative to the mission, with attempts to secure exceptions for scenarios such as a Chinese hypersonic missile or a drone swarm, which he said could not be relied on as permanent solutions. The Pentagon had designated Anthropic a supply chain risk, cutting off its defence work, and Anthropic has vowed to sue over that designation.

Michael noted that other AI companies accepted the Pentagon’s terms, while Anthropic resisted a push for “all lawful use” and for no mass surveillance of Americans. 7 March 2026.