Anthropic PBC v. United States Department of War
Issue
In *Anthropic PBC v. U.S. Department of War*, the government argues that a federal agency may validly exclude a domestic AI company from defense contracting based solely on that company's openly disclosed, commercially standard AI safety guardrails — characterizing those guardrails as a cognizable "supply-chain risk" under 41 U.S.C. § 4713. The case asks whether that statute's risk-designation authority extends to transparent, good-faith product-design choices by a domestic firm, and whether the Secretary's decision to bypass pre-designation notice on national-security urgency grounds was procedurally lawful and judicially reviewable.
What Happened
This is the government's merits brief, filed May 6, 2026 in the D.C. Circuit by DOJ Appellate Staff on behalf of the U.S. Department of War and Secretary Peter B. Hegseth, opposing Anthropic PBC's petition for review of a March 3, 2026 supply-chain risk designation that effectively excluded Anthropic from defense contracting. The brief defends the designation on four grounds: procedural compliance with § 4713's requirements, including permissible use of the statute's urgent-national-security bypass; a broad reading of § 4713(k)(6)'s "any person" language that reaches Anthropic's admitted model guardrails; strong judicial deference to the Secretary's predictive national-security judgment under *Holder v. Humanitarian Law Project* and *Trump v. Hawaii*; and the sufficiency of post-deprivation process to satisfy Fifth Amendment due process. The government further argues that Anthropic's failure to submit any response during the offered 30-day window renders any procedural sequencing defect harmless under *Shinseki v. Sanders*, and that the urgency determination itself is unreviewable by analogy to *Webster v. Doe*. Oral argument is scheduled for May 19, 2026.
Why It Matters
This case presents the first known attempt to deploy federal supply-chain risk management authority against a domestic AI company's safety architecture, and the court's resolution will define whether § 4713 can reach commercially transparent design choices or is confined to covert foreign-tampering threats — a question with immediate consequences for every AI vendor in the defense industrial base. The government's broad "any person" statutory construction, if accepted, would give national-security agencies sweeping authority to exclude AI firms whose models decline certain task categories, effectively converting safety-by-design into a procurement liability. Two doctrinal fault lines are particularly worth watching: whether the D.C. Circuit accepts the government's *Webster v. Doe* unreviewability analogy for urgency determinations — a position most observers regard as a significant stretch — and whether the court requires the government to address the due process implications of excluding a contractor from already-integrated existing systems, a gap the brief conspicuously leaves open.
Related Filings
Other proceedings in the same litigation tracked by this monitor.
How accurate was this summary?