At a hearing, two of three judges of a federal appeals court appeared skeptical of Anthropic's bid to block the DOD from designating it a supply-chain risk

A federal appeals court appeared skeptical of Anthropic's bid to block the Pentagon from declaring that the company poses …

Bloomberg 2026-05-19 Jen Judson

Context & Ripple Effects

Anthropic’s challenge has already produced conflicting procedural signals: a California judge granted a preliminary injunction in March, while the DC appeals court later declined to pause the Pentagon’s supply-chain-risk designation. The May hearing puts the dispute’s next meaningful constraint in the appellate process.

The case unfolds as the Pentagon expands its use of frontier AI, including a planned xAI integration into GenAI.mil. That makes the standards and process for excluding or restricting an AI provider consequential beyond Anthropic’s individual dispute.

First-order effects

Skepticism from two appellate judges reduces Anthropic’s apparent near-term leverage to stop the Pentagon’s designation through this appeal, though no final ruling is indicated.
The Pentagon retains a stronger immediate position in its dispute with Anthropic, preserving uncertainty around Anthropic’s ability to participate in or be treated as eligible for defense-related work while litigation continues.

Second-order effects

Other frontier-model vendors pursuing Pentagon work gain a clearer incentive to align their security, contracting, and governance practices with Defense Department expectations, because a supply-chain designation can become a practical commercial constraint even while contested.
The dispute may make the Pentagon’s vendor-selection process more important as it brings systems such as xAI’s into GenAI.mil: agency risk assessments could shape which providers can compete for sensitive deployments.

Third-order effects

If courts leave the Pentagon broad room to apply supply-chain-risk authorities to AI companies, national-security review could become a durable gatekeeper for the US government AI market alongside ordinary procurement rules.
The litigation also exposes a possible long-running tension between AI companies’ safety or policy positions and the government’s control over access to defense systems; its eventual legal resolution will help define how much procedural protection firms have when challenging exclusionary security designations.

The trend: Defense adoption of frontier AI is increasingly being governed not only by model capability, but by national-security risk determinations that can determine which vendors reach sensitive government deployments.

Discussion

@allinallnotbad Samuel Roland on x
Katsas getting down to brass tacks, what are the differences between the SCR designation and these less intrusive measure. Anthropic lawyer's response: 1. Permanent legal disbarment, blocks future modification 2. 4713 was meant to be a whole of government effort, creates risk of
@allinallnotbad Samuel Roland on x
Two orders that count as covered procurement orders, the March 3 and the March 6; both are evidence of covered procurement actions. Have to be honest here, can't tell whether Anthropic lawyer is claiming that the notices themselves or the downstream actions are the covered
@hadas_gold Hadas Gold on x
Anthropic and Pentagon are at the DC circuit court of appeals re: Supply Chain Risk designation. One of the judges, Henderson, seems incensed. “To me this is just a spectacular overreach by the department.” she also said “I see no evidence of that. I don't see that the department
@allinallnotbad Samuel Roland on x
Reminder before we start for those listening in that the panel draw was quite rough for Anthropic, so you should expect an uphill battle going in here.
@allinallnotbad Samuel Roland on x
Here we go. Immediately hot bench, already on to jurisdiction under 1327, question is “what is the covered procurement action here?” from Judge Rao. Anthropic responds that the written designation and notice from the DoW is authorizes all procurement actions, converts to
@allinallnotbad Samuel Roland on x
“To me this is just a spectacular overreach by the department” - Judge Henderson
@allinallnotbad Samuel Roland on x
Judge Henderson with an incredibly pro-Anthropic questions “I see no evidence of sabotage, malicious, etc. I don't see any evidence that the DoW has supported its designation that Anthropic is a bad actor?” (paraphrased here). WOW, did not expect that.

Chronicles