The US Department of War’s attempt to force AI company Anthropic to grant unrestricted military access to its model, Claude, was met with resistance. The Pentagon threatened to use the Defense Production Act, a law typically employed to secure physical goods or compel information sharing, not to alter AI safety features.