A California federal judge determined on Thursday evening that the Trump administration’s classification of Anthropic as a supply-chain risk was unlawful.
In her decision, U.S. District Judge Rita Lin stated that Defense Secretary Pete Hegseth’s branding of Anthropic as a national security threat constituted “unlawful retaliation,” infringing upon the First Amendment, and deemed the action “arbitrary and capricious.” Furthermore, Lin asserted that Anthropic was deprived of its Fifth Amendment right to due process.
Previously this year, both Hegseth and President Donald Trump declared Anthropic a supply-chain risk, instructing all federal entities, including non-defense agencies, to cease collaboration with the developer of Claude.
The disagreement originated from Anthropic’s firm stance on specific safety protocols, which would permit the Pentagon to utilize its models for completely autonomous weaponry and widespread surveillance of U.S. citizens. The Pentagon refuted claims that it would employ Anthropic’s models for anything other than legal applications, suggesting Anthropic might attempt to dictate the military’s usage of purchased models.
Judge Lin’s ruling indicated that the government’s “words and deeds corroborate that the contested actions stemmed from a desire to publicly chastise Anthropic for its ‘arrogance’ in criticizing the government.”
She highlighted the inconsistency between the supply-chain designation and other governmental actions, such as Hegseth’s proposal to invoke the Defense Production Act for Anthropic, “which would imply the company was vital to national security rather than a menace.” Additionally, she noted the Department of Defense’s continued pursuit of a contract with the firm and the government’s partnership on the company’s new cybersecurity model, Mythos.
Lin further clarified that Anthropic “unquestionably lacks” any surreptitious access to its technology once transferred to the DOD.
“While the Department of War is undeniably at liberty to choose its preferred AI vendor, the proof indicates that the extensive actions taken against Anthropic were unlawful and unfounded,” Lin stated. She further remarked, “The mere mention of national security does not grant unlimited power to penalize and retaliate against those who critique the government.”
An Anthropic representative conveyed to TechCrunch in a statement, “We applaud the court’s decision declaring this supply chain risk classification illegal. Our priority remains collaborating effectively with the government to leverage AI for national security, ensuring all Americans benefit from this technology.”
In March, Anthropic initiated two legal actions against the DOD, one in California and another in Washington, D.C., with the latter case still pending.
TechCrunch has sought a statement from the DOD.