Court Overturns Pentagon's Attempt to Blacklist Anthropic Over Defense Contract Dispute

A California federal judge has blocked the Pentagon's effort to exclude Anthropic from federal contracts, finding that the national security designation lacked legal basis and amounted to retaliatory action.

US District Judge Rita Lin's 59-page ruling on Thursday vacated Defense Secretary Pete Hegseth's February decision to label Anthropic a "supply-chain risk"—a move that would have made the company ineligible for government work. The court also struck down an accompanying measure that barred military contractors and suppliers from working with the firm.

Lin concluded that the broad sanctions imposed on Anthropic were "illegal and baseless," stating that "though the Department of War is undisputedly free to select the AI vendor of its choice, the evidence demonstrates that the broad measures imposed on Anthropic were illegal and baseless." The ruling additionally found that nine federal agencies, including the Treasury, State Department, and Department of Homeland Security, had improperly sanctioned the company.

The Contract Dispute#

The escalating tension began after news reports indicated that Claude, Anthropic's AI model, had been used in a US military operation targeting Venezuelan President Nicolás Maduro earlier this year. Following that disclosure, an Anthropic employee raised concerns with Pentagon officials about how the technology had been deployed.

Both parties had initially signed a $200 million agreement to integrate Claude into military applications. However, Anthropic sought contractual restrictions on model usage, refusing to support applications involving lethal autonomous weapons or mass surveillance systems. When Hegseth rejected these conditions—arguing the Pentagon retained full discretion over "all lawful use" once deployed—the parties reached an impasse.

The Pentagon's Rationale#

The Defense Department maintained that designating Anthropic a supply-chain risk was justified, claiming that permitting the company access to classified systems would create "unacceptable risk" if Anthropic could disable or modify its technology during wartime.

In response, Anthropic pursued two separate legal challenges—one in California federal court and a parallel case pending before the US Court of Appeals for the District of Columbia—alleging violations of its First and Fifth Amendment rights based on ideological grounds.

Implications and Next Steps#

The court emphasized the inconsistency of the Pentagon's position, noting that the government is currently engaged in discussions with Anthropic regarding collaboration on advanced models in sensitive national security contexts. A Pentagon spokesperson indicated the department intends to appeal. Anthropic has welcomed the ruling, stating it remains committed to productive collaboration with government agencies on AI applications for national defense.