A federal appeals court in Washington, D.C., is currently deliberating a significant legal challenge from Anthropic against the Defense Department's classification of the AI company as a supply-chain risk. This designation, typically applied to foreign adversaries, could deeply affect Anthropic's operations and its ties with the military.
On Tuesday, Anthropic and the Justice Department presented their arguments before a three-judge panel, which includes Judges Karen Henderson, Gregory Katsas, and Neomi Rao. Each side had 15 minutes to make its case before the judges issue a written ruling. Anthropic's legal team argues that the government's classification lacks factual support and violates constitutional procedures, claiming the label is both unjustified and unlawful.
The dispute began after negotiations between Anthropic and the DOD broke down. The department sought unrestricted access to Anthropic's AI models for various military applications, while Anthropic insisted on guarantees that its technology would not be used in autonomous weapons or for domestic surveillance. This impasse led Defense Secretary Pete Hegseth to blacklist the company, a decision that Anthropic's lawyers argue has inflicted significant reputational and operational damage.
In its pre-argument brief, the DOD expressed concerns that Anthropic's ability to "encode limitations" into its AI models presents an "untenable national-security risk." The document also noted that Secretary Hegseth believed the firm had undermined the trust essential for military collaboration by potentially controlling the application of its models, thereby imposing its ethical views on military operations.
Anthropic disputes these assertions, stating that the government's stance is baseless and lacks sufficient evidence to justify such a severe classification. The company claims its constitutional rights and established protocols were violated by this designation.
An earlier ruling from a court did not pause the DOD's classification but acknowledged that Anthropic could suffer irreparable harm during the ongoing legal proceedings. This recognition prompted a fast-tracked review of the case, highlighting the urgency involved.
At the same time, a parallel lawsuit is moving through federal court in San Francisco. This case emerged because the DOD's actions stemmed from two different designations, each requiring separate judicial oversight. In this instance, a judge provisionally sided with Anthropic, issuing an injunction that allows the company's models to remain available to federal agencies outside the Defense Department during the review process. "Nothing in the governing statute supports the Orwellian notion that an American company may be branded a potential adversary and saboteur of the U.S. for expressing disagreement with the government," the judge stated.
Anthropic's absence from a recent Pentagon announcement—where major tech companies, including Google, OpenAI, Microsoft, and Amazon Web Services, secured agreements to integrate their technologies into the Defense Department's classified networks—raises further concerns about its future with military contracts. While President Donald Trump suggested last month that a deal between the DOD and Anthropic remains possible, the ongoing legal challenges complicate any potential collaboration.
As the court deliberates, the outcome will have substantial implications not only for Anthropic but also for the future of AI development and its relationship with national security. The case underscores the delicate balance between promoting innovation in AI and ensuring that these technologies do not threaten national interests. The appeals court ruling is likely to set a precedent for how AI firms manage their relationships with government entities moving forward.
