Pentagon Wins the Appeal: Claude Stays Blacklisted
On September 25, 2026, the US Court of Appeals for the District of Columbia Circuit overturned the lower court's ruling and upheld the Pentagon's decision to designate Anthropic a supply chain risk, banning Claude from Department of War systems. In a 2 to 1 decision, Circuit Judges Gregory Katsas and Neomi Rao rejected every argument Anthropic raised: that the exclusion was arbitrary, that it exceeded the government's statutory authority, and that it violated the company's constitutional rights. Circuit Judge Karen LeCraft Henderson dissented. This result reverses the August 28 win Anthropic had secured before a district judge, who had called the designation illegal and baseless.
The Core of the Dispute: Who Controls the Model
The court's opinion, obtained by MaxAssistant, centers on a basic tension in enterprise AI deployment: who decides what a model is allowed to do once it is inside a government or corporate system. Anthropic builds restrictions directly into Claude that block lethal autonomous weapons use and mass domestic surveillance, regardless of what a government client requests. The Department of War argued that this is precisely the problem: a vendor that can alter a model's guardrails and weights after delivery introduces uncertainty into systems the military says must behave predictably. The court sided with the Pentagon, concluding that an incident in which Claude's restrictions reportedly interfered with an overseas military operation gave the government ample reason to doubt the model would perform as expected in a crisis.
A Long Fight That Keeps Shifting
The relationship between Anthropic and the Pentagon has deteriorated in public view since March 2026, when Defense Secretary Pete Hegseth formally labeled the company a supply chain risk under the Federal Acquisition Supply Chain Security Act, ordering Anthropic products removed from Department systems within 180 days and barring contractors from using them. Anthropic sued, won a temporary district court victory in August, and has now lost that ground on appeal. A separate, narrower case over a different rule used to justify the exclusion remains pending. Throughout the dispute, Anthropic CEO Dario Amodei has maintained that the company would offer a smooth transition to another AI provider if asked to leave, while insisting its safety guardrails are not up for negotiation, even with the US government as the client.
What This Means Beyond the Pentagon
For companies and agencies elsewhere that buy enterprise AI, including in Brazil, the ruling sets an important precedent: a government buyer can treat a vendor's own safety restrictions as a security liability rather than a selling point, and courts may back that framing even when the restrictions exist to prevent harm. This complicates the pitch that responsible AI design is purely a competitive advantage. It also signals that Anthropic's legal and commercial standing with large institutional buyers, particularly defense and government clients, now carries real uncertainty that was not fully priced in before this appeal. Microsoft has said it will keep embedding Anthropic's technology in products for clients other than the US Department of Defense, suggesting the fallout is being contained to military and defense-adjacent contracts for now.
The Bottom Line
MaxAssistant's read is that this ruling does not settle whether AI vendors should be allowed to enforce their own ethical limits inside government deployments, it simply confirms that, for now, the government gets to decide that question for its own systems. Anthropic's two red lines, no autonomous weapons and no mass surveillance, remain in place in its public commitments, but the company has just lost the legal fight to keep enforcing them inside the Department of War. Businesses evaluating any AI vendor's enterprise promises should treat built-in safety commitments as policy, not guarantee, since a buyer with enough leverage can simply route around a vendor that will not bend.
Sources
U.S. Court of Appeals for the D.C. Circuit: opinion in case 26-1049; CNBC: appeals court upholds Anthropic risk designation; ABC News: coverage of the September 25 ruling.