D.C. Circuit Declines to Stay Department of War’s “Supply-Chain Risk” Designation of Claude

From today’s order by Judges Karen LeCraft Henderson, Gregory Katsas, and Neomi Rao in Anthropic PBC v. U.S. Dep‘t of War:

Anthropic PBC develops Claude, a family of advanced artificial-intelligence models. In 2024, the Department of Defense (which now calls itself the Department of War) began using Claude in connection with various military operations.

But on March 3, 2026, Secretary of War Pete Hegseth determined that procuring AI goods or services from Anthropic presents a supply-chain risk to national security under 41 U.S.C. § 4713. The impetus for the determination was Anthropic’s refusal to contractually authorize the Department to use Claude for mass domestic surveillance or lethal autonomous warfare. As a result, the Department has canceled its contracts with Anthropic, begun to remove Claude from its systems, and prohibited its other contractors from using Anthropic as a subcontractor on work performed for the Department. The Department has not prohibited contractors from using Claude for work performed for entities other than the Department.

Anthropic seeks review of the Secretary’s determination under section 4713 to bar the company from providing goods or services to the Department. It claims that the determination was contrary to law, unconstitutional, and arbitrary. Anthropic seeks a stay pending review on the merits or, in the alternative, expedited consideration of the merits.

Four considerations govern whether Anthropic is entitled to the extraordinary remedy of a stay pending review: (1) whether it has made a “strong showing” that it is likely to succeed on the merits; (2) whether it will suffer irreparable harm without a stay; (3) whether a stay will injure the Department; and (4) whether the public interest supports a stay. Because the respondents are government agencies or officers, the third and fourth factors merge into a single inquiry.

Anthropic’s petition raises novel and difficult questions, including what counts as a supply-chain risk under section 4713 and what qualifies as an urgent national-security interest justifying the use of truncated statutory procedures. In addition, we must consider whether Anthropic’s petition targets a “covered procurement action” reviewable at this time under the governing judicial-review scheme, 41 U.S.C. § 1327(b). The parties vigorously contest many of these issues, and we have found no judicial precedent shedding much light on the questions presented. But we do not broach the merits at this time, for Anthropic has not shown that the balance of equities cuts in its favor.

We begin by acknowledging that Anthropic will likely suffer some degree of irreparable harm absent a stay. Anthropic casts its interests partly in constitutional terms, but those interests seem primarily financial in nature. Anthropic seeks a pre-deprivation hearing under the Fifth Amendment, yet such a hearing would be valuable to the company only as a means for preserving financially beneficial contracts. Anthropic also claims ongoing harms from retaliation for its constitutionally protected speech. But Anthropic does not show that its speech has been chilled during the pendency of this litigation, so these ongoing harms are also financial effects of the Department’s actions against the company.

The precise amount of Anthropic’s financial harm is not fully clear. Anthropic’s CEO, Dario Amodei, has publicly stated that the “vast majority” of Anthropic’s customers will be “unaffected” by the designation, since it “plainly applies only to the use of Claude by customers as a direct part of contracts with the Department of War, not all use of Claude by customers who have such contracts.” And some record evidence suggests that Anthropic has financially benefited from its refusal to accede to the Department’s request for permission to deploy Claude for all lawful uses. Add. 240 (Amodei statement to employees that “the general public or the media … see us as the heroes (we’re #2 in the App store now!)”); see also Scanlon, In Graphic Detail: How Anthropic’s Pentagon Refusal Is Paying Off in Downloads, Brand Trust, and Enterprise Deals, Digiday (Mar. 9, 2026), https://perma.cc/2B54-855B (“The $200 million [Anthropic] walked away from by refusing the Pentagon’s demands may turn out to be the best marketing spend in Silicon Valley for years.”).

Nonetheless, Anthropic has documented some potentially significant financial losses, particularly if other federal agencies follow the Department’s lead in removing Claude from their own supply chains. Absent any mechanism for Anthropic to recover these losses if it should prevail on the merits, these financial harms qualify as irreparable.

There are weighty governmental and public interests on the other side of the ledger. Most obviously, granting a stay would force the United States military to prolong its dealings with an unwanted vendor of critical AI services in the middle of a significant ongoing military conflict. As the Department explains, Anthropic has now conclusively barred uses that the Department recently deemed essential. See Add. 223 (Hegseth memo to senior Department leadership, dated Jan. 9, 2026, stating that “[t]he Department must also utilize models free from usage policy constraints that may limit lawful military applications”).

Moreover, the Department relies on Anthropic to provide regular updates to Claude, which contains built-in “safeguards” designed to prevent uses that Anthropic considers harmful. Furthermore, Anthropic and the Department recently disagreed about uses of Claude for military operations that the Department claimed were permitted under the existing usage policy. And the Department’s relationship with Anthropic has deteriorated to the extent that Anthropic’s CEO has publicly described the Department’s statements regarding the controversy as “completely false” and “just straight up lies.”

Under these circumstances, requiring the Department to prolong its use of Anthropic’s AI technology, whether directly or through contractors, strikes us as a substantial judicial imposition on military operations. And, of course, we do not lightly override the Department’s judgments on matters involving national security. See, e.g., Trump v. Hawaii (2018).

In our view, the equitable balance here cuts in favor of the government. On one side is a relatively contained risk of financial harm to a single private company. On the other side is judicial management of how, and through whom, the Department of War secures vital AI technology during an active military conflict. For that reason, we deny Anthropic’s motion for a stay pending review on the merits. { Our decision to deny a stay makes it unnecessary to address the government’s suggestion that the relevant judicial-review provision, 41 U.S.C. § 1327(b), deprives us of authority to grant one.}

Nonetheless, because Anthropic raises substantial challenges to the determination and will likely suffer some irreparable harm during the pendency of this litigation, we agree with Anthropic that substantial expedition is warranted.

The court has therefore set a quick briefing schedule, to be followed by oral argument for May 19; it instructs the parties to address:

• whether we have jurisdiction over Anthropic’s petition under 41 U.S.C. § 1327, which provides for review of “covered procurement actions” taken under 41 U.S.C. § 4713;
• whether the government has, through the Determination or Notice or otherwise, directed or taken specific covered procurement actions against Anthropic;
• whether, and if so how, Anthropic is able to affect the functioning of its artificial-intelligence models before or after the models, or updates to them, are delivered to the Department.

I’m trying to make certain how this interacts with the Mar. 26 preliminary injunction issued by Judge Rita Lin (N.D. Cal.) in the apparently parallel federal case filed by Anthropic in that court; see here for Monday’s status report from the government in that case. But I thought I’d note the D.C. Circuit action in any event.