US Appeals Court Upholds Pentagon Blacklisting of AI Firm Anthropic
A U.S. federal appeals court has declined to block a Defense Department order designating artificial intelligence startup Anthropic as a national security supply chain risk, upholding the military's ban on the company's Claude AI model. The ruling from the U.S. Court of Appeals for the District of Columbia Circuit marks a significant setback for the company as it navigates ongoing commercial disputes and prepares for a highly anticipated initial public offering.
The underlying conflict originated in February, when Anthropic rejected demands from Pentagon Secretary Pete Hegseth to remove safety restrictions preventing Claude from being deployed in fully autonomous lethal weapons and for the domestic surveillance of U.S. citizens. Following the refusal, the Defense Department classified Anthropic as a supply chain risk, effectively barring military contractors from integrating its technology and cutting off key defense contracts in an unprecedented application of the authority against a domestic firm.
Legal Arguments and Court Decision
In a 2-1 majority decision, the D.C. Circuit appeals panel concluded that the Pentagon possessed sufficient grounds to determine that utilizing Claude within military information systems presented a tangible national security risk. The court rejected Anthropic's central contention that the administration lacked statutory authority to impose the restrictions, ruling instead that the Pentagon acted because the company declined contract terms deemed vital to military command rather than in unlawful retaliation for its public policy positions.
Writing in dissent, one appellate judge argued that the Defense Department failed to adequately establish whether Anthropic genuinely met the formal legal criteria required for a supply chain risk designation. Despite the appellate setback, Anthropic pointed to a separate favorable ruling issued last month by a federal judge in San Francisco, which blocked a parallel government-wide ban after finding that the administration had retaliated against the firm over its stance on artificial intelligence safety.
Federal commitments to artificial intelligence software and services have experienced rapid growth, reaching a potential contract value share of $4.3 billion in 2024. Meanwhile, federal agencies frequently contract with commercial entities to acquire technical solutions as they seek to realize the benefits of artificial intelligence. Exclusion orders and supply chain risk determinations often involve shared patterns of concern regarding espionage, sabotage, or covert access to systems.
Government Response and Next Steps
Defense officials welcomed the D.C. Circuit decision, with Deputy Secretary of Defense Emil Michael posting on social media that the ruling demonstrated that private enterprises cannot insert their personal opinions into military chains of command. Meanwhile, Anthropic stated that it respectfully disagreed with the appellate outcome while remaining confident in its legal positioning.
Legal representatives for the company indicated that leadership is actively weighing further judicial review options, which could include petitioning the full D.C. Circuit for an en banc rehearing or escalating the dispute to the U.S. Supreme Court. Because the D.C. Circuit case and the California proceeding rely on distinct statutory authorities, the Pentagon's supply chain ban remains fully enforceable even as the broader government-wide restrictions face ongoing judicial scrutiny.