The executive branch has demonstrated it can repurpose national-security procurement law to punish AI companies for product decisions made on safety grounds. Section 3252's "supply chain risk" label was designed for foreign-adversary hardware — applying it to a domestic AI lab over guardrail policy is a structural expansion of executive power over private AI development that no legislation authorized Anthropic vs Pentagon: the AI safety lawsuit of 2026. Every AI lab with a federal contract now faces the same binary: comply with military use requirements or become, legally, a national…