DOJ defends Pentagon blacklist in Anthropic Claude lawsuit

President Trump signed Executive Order 13873 in May 2019 to block Chinese technology companies from U.S. defense supply chains. specifically Huawei and ZTE, which Congress later codified in Section 889 of the 2019 NDAA. The SCRM designation is the enforcement tool that mechanism created: contractors must certify they don't use flagged vendors. Defense Secretary Pete Hegseth applied it to Anthropic on March 3, 2026. the first time any administration has used a foreign-adversary exclusion mechanism against a domestic American company over a speech dispute about AI safety.

Anthropic won a $200 million Pentagon contract in July 2025 alongside Google, OpenAI, and xAI, all awarded $200 million each by the Pentagon''s chief digital officer for agentic AI prototyping on GenAI.mil. Negotiations over Claude''s deployment terms broke down because the Pentagon demanded "unfettered access" to Claude for all "lawful" uses, including fully autonomous weapons targeting and mass surveillance of Americans. Anthropic''s usage policies explicitly prohibit both. CEO Dario Amodei refused to remove them. The Pentagon cancelled the contract and issued the SCRM designation six months later.

Anthropic executives said in court filings the designation could cut 2026 revenue by billions and jeopardize hundreds of millions in existing private-sector contracts, as defense contractors were required to certify they don't use Claude in any Pentagon-related work.

Anthropic''s 48-page lawsuit, filed March 9, 2026 in the U.S. District Court for the Northern District of California, alleged First Amendment retaliationFirst Amendment RetaliationThe illegal use of government power to punish or deter someone for exercising their right to free speech.Key ConceptFirst Amendment RetaliationThe illegal use of government power to punish or deter someone for exercising their right to free speech.Open conceptBookmark, Fifth Amendment due processDue ProcessThe fundamental constitutional requirement that government follow fair procedures and apply laws reasonably to protect life, liberty, and property.Key ConceptDue ProcessThe fundamental constitutional requirement that government follow fair procedures and apply laws reasonably to protect life, liberty, and property.Open conceptBookmark violations, ultra vires executive action under the Youngstown framework, and that the presidential directive targeting Anthropic by name functions as an unlawful bill of attainder. The company named 18 federal agencies and 17 officials as defendants. A separate suit filed in D.C. federal court challenged the designation on administrative law grounds.

Legal experts cited by Reuters and CNBC called the bill of attainder argument particularly notable: Congress passed legislation in 1798 prohibiting exactly this kind of named-individual (or named-company) punishment by executive action, and the Supreme Court has applied it to corporate exclusions before.

U.S. District Judge Rita Lin granted Anthropic a preliminary injunction on March 26, 2026, writing that the designation constituted "classic illegal First Amendment retaliation" for the company''s publicly stated AI safety positions. To grant the injunction, Lin had to find that Anthropic was likely to succeed on the merits and would suffer irreparable harm without relief. Both findings are significant judicial conclusions: they mean a federal court assessed that the U.S. government probably violated the Constitution by using a national security procurement tool to punish a domestic company for speech.

The government filed an emergency appeal to the D.C. Circuit seeking to stay Lin''s injunction.

The D.C. Circuit denied Anthropic''s emergency bid to block the designation in April 2026, meaning the blacklist remained formally in effect even while the California district court injunction was nominally in place. Herbert Smith Freehills Kramer described the result as "two courts, two postures". the DC Circuit did not reach the constitutional merits, but its refusal to stay the designation created immediate compliance uncertainty for defense contractors.

Pentagon Chief Digital and AI Officer Michael Williams publicly stated Anthropic remained blacklisted despite Lin''s injunction, illustrating how executive branch agencies can create practical enforcement pressure even when a court order technically blocks the designation.

A three-judge D.C. Circuit panel, including Trump appointee Gregory Katsas and Judge Neomi Rao, heard nearly two hours of oral arguments on May 19, 2026. Courthouse News reported the panel appeared likely to find the Pentagon overstepped its authority. judges questioned whether the SCRM mechanism was designed to address this kind of domestic procurement dispute and pressed government counsel on the constitutional implications.

Anthropic's counsel framed the issue as a contract dispute in which the Pentagon used a narrow national security tool as leverage. The government countered that private companies can''t dictate how the military uses technology in warfare. The panel took the matter under advisement with no ruling issued as of June 2026.

A coalition of tech industry groups including the Information Technology Industry Council (ITI), TechNet, CCIA, and SIIA filed an amicus brief supporting Anthropic in the D.C. Circuit. They argued the designation disrupts established government contractingGovernment ContractingPrivate companies doing public work under contract.Key ConceptGovernment ContractingPrivate companies doing public work under contract.Open conceptBookmark procedures and creates immediate business uncertainty for the entire U.S. tech sector competing with China in defense AI. Separately, dozens of scientists at OpenAI and Google DeepMind filed personal amicus briefs arguing the designation would chill AI safety research.

CCIA stated the Pentagon''s actions "create substantial business uncertainty at a time when US companies are competing with global counterparts to lead in AI."

The Trump administration filed a formal denial of unlawful retaliation on June 9, 2026, with DOJ arguing that private companies can''t set the terms for how the U.S. military uses their technology. The government''s position: all proposed military uses of Claude would be "lawful," so Anthropic''s restrictions aren''t AI safety policy. they''re a commercial veto on legitimate defense operations.

This legal position directly conflicts with Judge Lin''s preliminary injunction finding. If the D.C. Circuit sides with the government, every AI company that publishes ethics guidelines for government use of its models faces the same risk of exclusion for maintaining those guidelines. By June 2026, DOD had formalized contracts with OpenAI, Google, and Microsoft. building a multi-vendor AI strategy that doesn''t depend on Anthropic.

Lawfare''s analysis of the May 19 oral arguments identified three questions the D.C. Circuit directed the parties to brief: whether the circuit has jurisdiction to review the SCRM designation, the standard of review that applies, and the conceptual difficulty of issuing a static ruling for a rapidly evolving AI contract relationship. Those questions signal the court may find a procedural exit that avoids the First Amendment ruling entirely. potentially sending the case back to the district court rather than resolving the constitutional question.