A federal judge in San Francisco, U.S. District Judge Rita Lin, ruled that the Department of Defense acted illegally when it designated Anthropic a supply chain risk, a label that restricted the AI company from certain military contracts. Lin found the designation violated the First Amendment, writing that the Pentagon's invocation of national security was not a blank check to punish and retaliate against government critics, and calling the action illegal and baseless. The dispute traces back to failed negotiations between Anthropic and Defense Secretary Pete Hegseth's Pentagon over unrestricted use of Claude models; after talks broke down, the department blacklisted the company, which Anthropic argued was retaliation for its public positions that current AI models are not reliable enough to be trusted with autonomous weapons decisions and that it would not support domestic mass-surveillance uses of its technology. For developers and companies building on foundation models, the ruling matters beyond the Anthropic-specific dispute: it is one of the first court tests of whether the federal government can use supply-chain risk designations, normally reserved for security threats like compromised hardware vendors, as a lever to pressure AI labs on how their models may be used by military and intelligence customers. A win for the Pentagon would have set a precedent letting agencies coerce AI vendors on usage policy through procurement blacklisting rather than negotiated contract terms, so the ruling preserves labs' ability to set and enforce their own usage policies even when a major government customer objects to them.