Court Sides With Pentagon in AI Firm Anthropic Case
▼ Summary
– A U.S. appeals court in Washington, D.C., ruled that Anthropic did not meet the requirements to lift a Pentagon supply-chain-risk designation, creating a conflict with a previous San Francisco court order.
– The case is unprecedented, with the appellate panel stating that removing the designation would force the military to prolong dealings with an unwanted AI vendor during an ongoing conflict.
– The conflicting rulings stem from two different supply-chain laws, with the San Francisco judge finding the Department of Defense likely acted in bad faith against Anthropic over its usage limits.
– Anthropic argues it is being illegally punished for stating its AI lacks the accuracy for sensitive military operations, like unsupervised drone strikes, and claims the designation has cost it business.
– The legal battle tests executive branch power over tech companies, with final decisions months away and oral arguments in the Washington case scheduled for May 19.
A federal appeals court has denied Anthropic’s request to suspend a Pentagon supply-chain risk designation, creating a direct conflict with a lower court’s ruling last month. The D.C. Circuit found that the AI company failed to meet the strict legal requirements for an injunction, allowing the contested national security label to remain in effect for now. This unprecedented legal clash centers on the government’s authority to restrict a domestic technology firm under laws typically aimed at foreign entities.
In its Wednesday decision, a three-judge panel emphasized the potential disruption to military operations. The judges acknowledged that Anthropic might face financial injury but stated that granting a stay would force the U. S. military to continue working with an “unwanted vendor” during an active conflict. The ruling explicitly declined to “lightly override” military judgments on national security concerns, prioritizing operational continuity over the company’s immediate grievance.
This outcome starkly contrasts with the preliminary injunction granted by a San Francisco judge in March. That court found the Department of Defense likely acted in “bad faith,” motivated by frustration over Anthropic’s public stance on ethical AI use. Following that order, the Trump administration temporarily restored federal access to Anthropic’s Claude AI tools. The conflicting opinions stem from two separate lawsuits, each challenging one of two similar supply-chain laws used to sanction the company.
Anthropic maintains its position that the designations are unlawful. A company spokesperson stated they are grateful the D. C. court recognized the need for a swift resolution and expressed confidence that the courts will ultimately rule in their favor. The company argues it is being penalized for its own assessment that its Claude AI lacks the requisite accuracy for sensitive, life-and-death military applications, such as autonomous drone strikes.
The acting attorney general, Todd Blanche, hailed the appellate decision as a “resounding victory for military readiness.” In a public statement, he asserted that military authority resides with the Commander-in-Chief and the Department of War, not a technology company, and that the Pentagon requires full access to AI models integrated into sensitive systems.
This legal battle tests the limits of executive branch power over technology firms, especially as the Pentagon accelerates AI deployment. Observers note that while Anthropic may have a strong legal argument, courts are often reluctant to second-guess the White House on security matters. Some experts warn that the government’s actions could have a chilling effect on essential professional debate regarding AI system performance and limitations.
The immediate business impact for Anthropic is significant. The designation bars the Pentagon and its contractors from using Claude AI for military projects, a restriction the company claims has already cost it revenue. With the Trump administration in power, regaining its former foothold in the federal marketplace remains a formidable challenge.
Final rulings in the dual lawsuits are likely months away, with oral arguments in the D. C. case scheduled for May 19. Key operational details remain undisclosed, including the specific military uses for Claude or the progress in transitioning to alternative AI tools from providers like Google DeepMind or OpenAI. The military has stated it has taken precautions to prevent any potential sabotage of its AI tools during this period of transition.
(Source: Wired)
