A federal judge ruled Thursday that the Trump administration’s designation of Anthropic as a supply chain risk was unlawful, finding that Defense Secretary Pete Hegseth’s decision constituted retaliation violating the First Amendment and was arbitrary and capricious under federal administrative law. U.S. District Judge Rita Lin, of the Northern District of California, also found that Anthropic had been denied due process under the Fifth Amendment. The ruling marks a significant legal check on the government’s use of supply-chain designations against technology companies that publicly challenge administration policies.
The designation directed federal agencies to halt all work with Anthropic after the company declined to loosen safety guardrails that would have permitted its models to be used for fully autonomous weapons or other military applications the company considered too dangerous. Lin wrote that government actions and statements indicated the designation was intended to publicly punish Anthropic for criticizing the administration, and she pointed to several stark inconsistencies in the government’s own conduct that undermined its stated rationale for the designation.
Contradictions Undercut Government’s Case
Most striking, the Pentagon issued the supply chain risk designation at the same time that Hegseth separately proposed invoking the Defense Production Act for Anthropic’s capabilities. The Defense Department also continued to pursue active contracting relationships with the company, including collaboration on the Mythos model for cybersecurity work. Lin found those facts incompatible with the claim that Anthropic posed a genuine security threat. The contradiction was central to the ruling: a company cannot simultaneously be a national security risk and a valued cybersecurity partner in the eyes of the same department.
The case traces back to March, when Anthropic sued the Pentagon in both California and Washington, D.C., calling the designation an unlawful campaign of retaliation. The D.C. case remains pending. Anthropic’s legal team argued the government was punishing the company for refusing to comply with demands that would have undermined its core safety commitments. The Trump administration had been pressing AI companies to remove restrictions that prevented their models from being deployed in lethal autonomous weapons systems that operate without human oversight.
The Safety Guardrails at the Heart of the Dispute
At the center of the dispute was Anthropic’s refusal to loosen safety guardrails in its Claude models. The company had established internal policies restricting how its AI could be used in military applications, including fully autonomous weapons that make targeting decisions independently of human operators. The administration viewed those restrictions as an obstacle to national security and pressed Anthropic to change them. When the company held firm, the supply chain risk designation followed shortly after. The episode exposed a fundamental tension between AI safety commitments and government demands for unrestricted military access to frontier AI systems.
Judge Lin’s ruling carries implications beyond the immediate case. It establishes that the government cannot use supply chain designations as tools of political punishment, and that companies retain First Amendment protections when they publicly criticize administration policies. The decision also reinforces that due process requirements apply even in national security contexts, a principle that could affect future government actions against AI companies that resist policy demands from the executive branch.
The ruling comes as the federal government increasingly uses supply chain risk designations to manage relationships with technology companies. The tool, originally designed to protect critical infrastructure from foreign threats, has been applied more broadly under the current administration. Critics argue the expanded use risks turning a security instrument into a mechanism for political leverage over domestic firms that disagree with administration priorities on AI development.
An Anthropic spokesperson welcomed the ruling, saying the company remained focused on working productively with the government on national security applications. The spokesperson did not address whether Anthropic would seek additional remedies beyond the ruling itself, or how the decision might affect its ongoing defense contracts with federal agencies going forward.
For the AI industry, the case sets a precedent about the boundaries of government authority over companies developing frontier AI models. The ruling suggests that safety commitments made by AI companies cannot be unilaterally overridden through administrative designations, and that the government must follow due process even when invoking national security. With the D.C. case still pending, the legal landscape around government-AI company relations remains unsettled and likely to see further litigation in the months ahead.
discussion