Jump to content

Anthropic PBC v U.S. Department of War

From AI Law Wiki

AI-generated text. This page was generated using artificial intelligence.

Case Information
Case name Anthropic PBC v. U.S. Department of War[1]
Court U.S. District Court for the Northern District of California[1]; separate procurement review in the U.S. Court of Appeals for the D.C. Circuit[2]
Docket 3:26-cv-01996-RFL[1]; D.C. Cir. Nos. 26-1049 and 26-1162[2]
Filed March 9, 2026[1]
Judge(s) Rita F. Lin[1]
Plaintiffs Anthropic PBC[1]
Defendants U.S. Department of War and named federal agencies and officials[1]
Case type Civil constitutional and administrative-law action[1]
Claims / issues First Amendment retaliation, due process and statutory challenges to supply-chain-risk restrictions[1]
Status October 26, 2026: federal defendants appealed the California merits judgment and final relief to the Ninth Circuit. The separate D.C. Circuit rehearing proceeding concerns the § 4713 procurement exclusion.[3]

Anthropic PBC v. U.S. Department of War (Case No. 3:26-cv-01996-RFL) is a federal lawsuit in which AI company Anthropic PBC challenges the Trump administration's designation of the company as a supply-chain risk to national security, alleging that the designation was retaliatory punishment for Anthropic's refusal to remove AI safety restrictions from its Claude model. On March 26, 2026, U.S. District Judge Rita F. Lin granted Anthropic a preliminary injunction blocking the government-wide ban on Anthropic's technology, finding the company likely to succeed on First Amendment retaliation and Fifth Amendment due process claims.[4][5][6] On August 27, 2026, Judge Lin ruled for Anthropic on the merits, finding that the challenged measures violated the First Amendment and the Fifth Amendment’s Due Process Clause.[7][8]

Case Background

On March 9, 2026, Anthropic filed suit against the U.S. Department of War, Secretary Pete Hegseth, and 17 federal agencies in the U.S. District Court for the Northern District of California, challenging a presidential directive and the Department's designation of Anthropic as a supply-chain risk under 41 U.S.C. § 4713 and 10 U.S.C. § 3252. The designation would have barred Anthropic and its affiliates from supplying products or services to the Department of War and, under a government-wide directive, effectively prevented any federal agency from using Anthropic's technology.[4][6]

Anthropic alleged the designation was issued after the company publicly refused to remove two usage restrictions from its Claude AI system: prohibitions on lethal autonomous warfare and mass surveillance of Americans. The company argued the government's actions constituted impermissible retaliation against Anthropic for exercising its First Amendment rights to speak out on matters of public concern and to set terms for the use of its own products.[5][9]

Claims

Anthropic's complaint asserted five counts:[4]

  1. Administrative Procedure Act violations — The supply-chain risk designation was arbitrary, capricious, and not in accordance with law
  2. First Amendment retaliation — The government punished Anthropic for protected speech (refusing to remove AI safety restrictions)
  3. Ultra vires executive action — The President exceeded statutory authority in directing a government-wide ban
  4. Fifth Amendment due process — Anthropic was deprived of liberty and property interests without due process
  5. APA sanction violations — The designation constituted an unlawful sanction under APA

Preliminary Injunction (March 26, 2026)

On March 26, 2026, Judge Rita F. Lin granted Anthropic's motion for a preliminary injunction, issuing the following key findings:[5]

  • Anthropic demonstrated a high likelihood of success on its First Amendment retaliation claim, having made a prima facie case that the government's actions were motivated by Anthropic's protected speech
  • Anthropic was likely to succeed on its Fifth Amendment due process claim
  • The government failed to prove that Anthropic's conduct qualifies as a supply-chain risk under the relevant statutes
  • The court found evidence that the designation appeared pretextual — designed to punish Anthropic for criticizing the DOD's contracting position rather than addressing genuine national security concerns

Judge Lin granted the government a seven-day administrative stay to seek an emergency, expedited stay from the Court of Appeals. The GSA restored Anthropic's status by early April 2026, and a government status report was filed on April 6, 2026 (Dkt. No. 146).[10]

A separate D.C. Circuit challenge to procurement actions under 41 U.S.C. § 4713 was decided on September 25, 2026, when a divided panel denied Anthropic’s petitions for review.[11]

D.C. Circuit supply-chain decision (September 25, 2026)

In a separate proceeding, Anthropic PBC v. U.S. Department of War, Nos. 26-1049 and 26-1162, the U.S. Court of Appeals for the D.C. Circuit denied Anthropic’s petitions for review of the Pentagon’s exclusion of Claude from its supply chain under the Federal Acquisition Supply Chain Security Act, 41 U.S.C. § 4713. Judge Gregory Katsas wrote for a 2–1 majority joined by Judge Neomi Rao; Judge Karen LeCraft Henderson dissented.[11]

The majority held that the department reasonably found a covered supply-chain risk in Anthropic’s ability to build usage restrictions into Claude and to encode restrictions in later versions supplied to defense contractors. It accepted the department’s concern that model restrictions could interrupt military uses amid a dispute over Anthropic’s contractual limits on lethal autonomous warfare and domestic surveillance. The majority also upheld the finding that less intrusive measures were not reasonably available. It found no prejudicial error from the timing of notice because Anthropic received a later opportunity to contest the exclusion, and rejected the company’s due-process and First Amendment challenges. On the First Amendment claim, the court distinguished Anthropic’s protected public advocacy from its refusal to accept a contract term the department considered essential.[11]

Judge Henderson’s dissent read the statute’s definition of “supply chain risk” as requiring intentionally subversive conduct, such as sabotage or deceptive manipulation. On her reading, Anthropic’s disclosed contractual and model safeguards did not meet that definition. The majority held that § 4713’s definition reaches manipulation of a covered system without a bad-motive requirement.[11]

The panel distinguished this § 4713 procurement proceeding from Judge Lin’s August 27 ruling concerning the designation under 10 U.S.C. § 3252 and broader government measures. The majority said § 3252 uses a narrower definition tied to an “adversary”; the California ruling therefore did not control its analysis of § 4713. The district court ruling remains a separate judgment, and the D.C. Circuit opinion resolves the petitions before that court.[11] See September 25 news coverage.

October 2026 rehearing proceedings

Anthropic petitioned for panel rehearing or rehearing en banc in the separate D.C. Circuit procurement proceeding on October 13, 2026. On October 15, the court directed a response to the petition for panel rehearing within 15 days, limited to 3,900 words, and barred a reply absent a further court order. On October 19, the government requested a 14-day extension from October 30 to November 13 because of counsel’s other appellate commitments; Anthropic took no position. The reviewed docket through October 21 does not show an order granting that extension or granting rehearing.[12] [13][2]

On October 20, 17 former senior national-security government officials submitted an amicus brief supporting rehearing. Signatories include Avril Haines, Susan Rice and Jake Sullivan. They argue that the designation preceded the required administrative process, that later explanations did not justify the earlier decision, and that the department failed to weigh the national-security costs of excluding Anthropic or discouraging AI safeguards. They also argue that the designation punished protected speech and that national-security deference does not displace independent First Amendment review. These are the amici’s arguments, rather than new judicial findings.[14]

The brief supports reconsideration of the September 25 panel decision under 41 U.S.C. § 4713. It identifies the California action under 10 U.S.C. § 3252 as related but separate. Submission of the brief does not vacate the panel decision, grant rehearing or change the California judgment.[14][2] See October 21 news coverage.

Amicus Briefs

The case attracted significant amicus support. Multiple organizations filed briefs supporting Anthropic, including:[9][15][16]

  • Cato Institute — Argued the government's retaliation violates core First Amendment principles
  • Society for the Rule of Law — Former national security officials warned the designation undermines national security by politicizing procurement
  • Electronic Frontier Foundation (EFF) — Emphasized the chilling effect on AI safety research
  • Industry trade associations (ITI and others) — Warned of broad harmful impact on the technology sector
  • Yale Law School Rule of Law Clinic — Argued the executive action exceeds statutory authority


On June 5, 2026, federal attorneys disclosed that the Department of Defense had denied Anthropic's request to reconsider the supply-chain risk designation.[17] Defense Secretary Pete Hegseth confirmed that the Pentagon would not reverse the designation, preserving the dispute for further court proceedings.[18]

Merits Ruling (August 27, 2026)

On August 27, 2026, Judge Lin ruled for Anthropic on the merits, finding that the government retaliated against the company for constitutionally protected criticism and violated the Fifth Amendment’s Due Process Clause.[7][8] The court described the broad measures as “illegal and baseless,” finding no articulable basis to believe Anthropic would sabotage its model.[7][8]

Significance

The case is one of the first major legal confrontations between an AI company and the U.S. government over AI safety restrictions and government retaliation. Its implications include:

  • AI safety vs. government mandates: Tests whether the government can punish AI companies for maintaining safety guardrails that conflict with government preferences
  • First Amendment protections for AI companies: Establishes that AI companies have constitutional protections when setting terms for the use of their products
  • Supply-chain risk designation limits: Clarifies that government procurement restrictions must be based on genuine security concerns, not retaliation
  • Chilling effect on AI safety: The ruling bars the challenged retaliation against Anthropic’s public advocacy for AI safety restrictions.[8]

Current Status

The Northern District of California case was decided in Anthropic’s favor on August 27, 2026.[7][8] On September 25, 2026, the D.C. Circuit denied Anthropic’s separate petitions challenging its exclusion from the Pentagon supply chain under 41 U.S.C. § 4713.[11] Anthropic’s October 13 rehearing request is followed by an October 15 response order and October 20 amicus submissions; no rehearing grant appears in the reviewed docket through October 21.[2]

October 26, 2026: government appeals California judgment

The federal defendants, represented by the U.S. Department of Justice, filed a notice of appeal to the Ninth Circuit on October 26. It identifies the order on cross-motions for summary judgment (ECF 250), order of final relief (ECF 251) and final judgment (ECF 252) as the decisions appealed. The notice initiates appellate review; it does not itself reverse the judgment or grant a stay.[3]

This appeal is separate from Anthropic’s October 13 request for rehearing of the D.C. Circuit’s September 25 decision under 41 U.S.C. § 4713.


See Also

References

  1. ↑ 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 Court filing / case record, reviewed September 7, 2026
  2. ↑ 2.0 2.1 2.2 2.3 2.4 D.C. Circuit public RECAP docket, Nos. 26-1049 and 26-1162, entries through October 21, 2026, reviewed October 21.
  3. ↑ 3.0 3.1 Defendants, notice of appeal, ECF 253, October 26, 2026, pp. 1–2
  4. ↑ 4.0 4.1 4.2 Clearinghouse, Anthropic PBC v. U.S. Department of War
  5. ↑ 5.0 5.1 5.2 Bloomberg Law, Preliminary Injunction Order, March 26, 2026
  6. ↑ 6.0 6.1 U.S. District Court, N.D. Cal., Case 3:26-cv-01996
  7. ↑ 7.0 7.1 7.2 7.3 Associated Press, “Judge rules in favor of Anthropic in case against Pentagon,” August 28, 2026
  8. ↑ 8.0 8.1 8.2 8.3 8.4 The Hill, “Judge rules Pentagon’s supply chain risk designation for Anthropic was illegal,” August 27, 2026
  9. ↑ 9.0 9.1 Cato Institute, Amicus Brief in Support of Anthropic, March 2026
  10. ↑ GSA Statement on Anthropic Preliminary Injunction, April 3, 2026
  11. ↑ 11.0 11.1 11.2 11.3 11.4 11.5 U.S. Court of Appeals for the D.C. Circuit, Anthropic PBC v. U.S. Department of War, Nos. 26-1049 and 26-1162, opinion and dissent, September 25, 2026, pp. 4–5, 17–18, 29–31, 32–43, 44–51.
  12. ↑ D.C. Circuit, order directing response to the petition for panel rehearing, document 2198350, October 15, 2026, p. 1.
  13. ↑ Respondents, motion for a 14-day extension to respond to the rehearing petition, document 2199042, October 19, 2026, pp. 1–6.
  14. ↑ 14.0 14.1 Former Senior National Security Government Officials, amicus brief supporting panel rehearing or rehearing en banc, D.C. Cir. Nos. 26-1049 and 26-1162, October 20, 2026, 26 PDF pages (argument pp. 1–11; Appendix A pp. 13–15).
  15. ↑ Society for the Rule of Law, Amicus Brief in Support of Anthropic
  16. ↑ Yale Law School, Amicus Brief in Support of Anthropic
  17. ↑ GovInfoSecurity, "DoD Says No to Anthropic Request for Reversing Blacklisting" (June 5, 2026).
  18. ↑ Benzinga, "Pete Hegseth Confirms Pentagon Won't Back Down On Anthropic Risk Designation As Legal Fight Continues" (June 5, 2026).
Retrieved from "https://ailawwiki.com/index.php?title=Anthropic_PBC_v_U.S._Department_of_War&oldid=7462"