A federal judge in California ruled that the Trump administration’s national-security designation barring Anthropic from certain U.S. government work was unlawful retaliation for constitutionally protected speech, the New York Times and Bloomberg reported. The government must lift the supply-chain risk label that had frozen Anthropic out of Pentagon-related contracts.
The dispute grew out of Anthropic’s public criticism of Defense Department AI plans. Officials tagged the company as a supply-chain risk — a designation that functions in practice as a ban on selling into large parts of the federal stack.
The ruling does not decide whether Anthropic’s models should be used in classified settings. It decides that agencies cannot use security designations as a shortcut to punish labs that criticize policy.
Restoring eligibility for federal work removes one overhang as Anthropic prepares a larger capital-markets story. It does not end the fight over how far the executive branch can go in gating frontier-model releases.
Key takeaway
The court drew a line between genuine supply-chain risk review and using that label to retaliate against an AI lab’s public speech.
Photo: Aerial view of the Pentagon, January 2008 (Wikimedia Commons / U.S. government work). Sources: The New York Times · Bloomberg · Reuters. Reporting compiled August 31, 2026.
