A US federal appeals court has upheld the Pentagon’s blacklisting of Anthropic from military contracts, backing the Defense Department’s argument that the AI company’s safety restrictions could interfere with battlefield operations.
The 2-1 decision by the US Court of Appeals in Washington came in Anthropic’s challenge to its March designation as a national-security supply-chain risk. The company says the designation has cost it billions of dollars in lost business and damaged its reputation ahead of a planned initial public offering.
The ruling extends a months-long dispute over how much control AI companies can retain over the military use of their models.
Proposal for Conducting Cyber Crisis Drill, Tabletop Exercise (TTEx) & CCMP Readiness Exercise
Court backs Pentagon concerns over battlefield reliability
The majority said it was reasonable for the Pentagon to act after Anthropic refused to permit Claude to be used for autonomous weapons or mass surveillance.
Judge Gregory Katsas wrote that the Defense Department had raised a serious concern that overly constrained AI systems could shut down unexpectedly and cause important military operations to fail.
The court also rejected Anthropic’s argument that the blacklist was retaliation for the company’s views on AI safety and ethics.
That finding is significant because Anthropic has repeatedly argued that the government punished it for publicly maintaining safeguards around military use.
The appeals court instead treated the dispute as one involving procurement and national-security risk.
Anthropic opposed autonomous weapons and mass surveillance
The confrontation began after the Defense Department pressed Anthropic to loosen restrictions on how Claude could be used.
Anthropic said it would not remove safeguards preventing use in autonomous weapons systems or mass domestic surveillance.
The company has argued that current AI systems are not reliable enough to be trusted with fully autonomous lethal decisions.
It has also said that unrestricted domestic surveillance would raise fundamental-rights concerns.
Defense Secretary Pete Hegseth took the opposite view, arguing that contractual limits imposed by an AI supplier could undermine military readiness if the technology became unavailable or refused to perform during an operation.
That disagreement eventually moved beyond contract negotiations into a broader legal and political conflict.
Pentagon used two separate legal routes against company
The Defense Department designated Anthropic a supply-chain risk under two different legal authorities after negotiations between the company and the Pentagon failed.
The government also moved to prevent military contractors from doing business with Anthropic.
Anthropic challenged those actions in court.
The legal picture became complicated because separate courts reached different conclusions under the two legal frameworks.
A federal judge in San Francisco last month struck down a parallel designation and blocked a broader government ban, finding that the administration had unlawfully retaliated against Anthropic over its AI-safety views.
Friday’s Washington appeals-court ruling concerns the other legal route and leaves that Pentagon blacklist in place.
Split rulings mean legal fight is not over
The September 25 decision does not end the dispute.
Anthropic said it respectfully disagrees with the ruling and is considering its options, including asking the full DC Circuit to rehear the case.
That process, known as en banc review, would allow a larger group of judges on the appeals court to reconsider the three-judge panel’s decision.
Anthropic could also potentially seek review by the US Supreme Court later, depending on how the litigation develops.
For now, however, the Pentagon’s supply-chain-risk designation remains effective under the ruling upheld on Friday.
Government ties improved even as blacklist remained
The legal battle has unfolded alongside a more complicated relationship between Anthropic and the Trump administration.
Reuters reported earlier this year that ties between the company and parts of the White House had begun improving despite the Pentagon dispute. Anthropic executives continued discussions with government officials as the company prepared for a possible IPO.
Some government agencies also continued using Anthropic tools even while the Pentagon restrictions remained in place.
That shows the dispute is not a complete government-wide rejection of the company.
It is more specifically focused on whether Anthropic’s contractual safety limits are compatible with the Defense Department’s requirements.
Case raises wider question over control of military AI
The broader issue reaches beyond Anthropic.
AI companies increasingly provide models that governments may want to use for intelligence analysis, cyber operations, targeting support and other military functions.
That creates a difficult question: who gets the final say over how a commercial AI model can be used once it is sold to the government?
AI developers may want contractual limits designed to prevent uses they consider unsafe or unethical.
Military agencies, on the other hand, may argue that critical systems cannot depend on private companies retaining the ability to restrict functions during a conflict.
The Anthropic-Pentagon fight is one of the clearest legal tests of that tension so far.
Friday’s ruling gives the Defense Department more room to treat supplier-imposed restrictions as a national-security risk, at least under the authority examined by the DC Circuit.
What this means for you: The ruling does not mean Anthropic’s safety concerns were disproved. It means this appeals court accepted the Pentagon’s argument that those restrictions could create unacceptable operational risk for military use. The legal fight may continue through a full-court rehearing or further appeal.
Follow for daily updates on cybercrime, corporate fraud, DFIR, hacking, investigations, and digital forensics
