Anthropic Loses Appeal Against Pentagon Blacklisting

A U.S. federal appeals court in Washington, D.C. on April 8 denied Anthropic’s emergency request to stay a Pentagon ‘‘supply‑chain risk’’ designation while the company pursues judicial review. The decision leaves the Defense Department’s label available to govern Pentagon procurement and contractor use pending the underlying litigation.

Technical and legal context The Department of Defense designated Anthropic a supply‑chain risk in early March, a move the Pentagon says stems from concerns about how the company’s Claude models could be used in ways that threaten national security. The designation requires military components and defense contractors to certify they do not use Anthropic’s Claude models. Anthropic argues the designation is arbitrary, retaliatory and procedurally flawed and filed separate lawsuits in multiple courts challenging the action and related directives from the administration.

• •The D.C. appeals court concluded the equitable balance favors the government, reasoning that the government’s need to ‘‘secure vital AI technology during an active military conflict’’ outweighs the comparatively contained financial harm to one private company. The court acknowledged Anthropic may suffer ‘‘some degree of irreparable harm’’ but characterized that harm as primarily financial. (CNBC)
• •Reuters reported Anthropic told the court more than 100 enterprise customers contacted the company about the designation and estimated the action could cost ‘‘hundreds of millions, or even multiple billions’’ in lost 2026 revenue. (Reuters)
• •A separate San Francisco federal judge (referred to in filings as Judge Lin) granted a preliminary injunction in a related case that temporarily blocks a broader federal ban on the use of Claude; the DOJ has signaled appeals in that matter and the Ninth Circuit set a briefing schedule. (AP)
• •The dispute intensified after negotiations over a defense contract broke down when Anthropic sought contractual guardrails limiting use of its models for certain surveillance or domestic operations; the Pentagon pushed back that it must be free to use acquired AI tools ‘‘in any way it deems lawful.’’ Senior DoD officials, including Defense Secretary Pete Hegseth, publicly supported the designation. (AP, Military.com)

This litigation sits at the intersection of AI governance, procurement policy, and operational security. For ML engineers and program managers supplying models to government or regulated industries, the case demonstrates (1) how vendor decisions around model safety guardrails and usage restrictions can trigger procurement risk, (2) the practical effect of an administrative ‘‘supply‑chain risk’’ label on commercial revenue and customer behavior, and (3) the legal levers governments can use to shape vendor behavior when national‑security concerns are asserted. Large enterprises that contract with the DoD or its contractors must track supplier certifications and may need contingency plans for rapid vendor substitution.

• •Appeals timeline: Anthropic’s D.C. Circuit challenge and the DOJ’s appeals of the San Francisco preliminary injunction are the next legal milestones; the Ninth Circuit set an April 30 deadline for filings in one matter. Outcomes will set precedent on the scope of administrative labeling and the courts’ deference to national‑security judgments about AI vendors. (AP)
• •Commercial impact signals: Monitor enterprise customer disclosures and Anthropic’s revenue guidance for evidence of contract losses or contract churn tied to the designation. (Reuters)
• •Procurement policy evolution: Watch whether DoD refines criteria or processes for supply‑chain risk designations and whether Congress or oversight bodies press for clearer standards for AI vendor blacklisting.