TL;DR
A federal appeals court ruled that the Pentagon lawfully blacklisted Anthropic as a national security supply-chain risk after the company refused to let its Claude AI perform mass domestic surveillance or autonomous weapons tasks, creating a direct circuit split with a California court that called the same action unconstitutional.
What happened
- The U.S. Court of Appeals for the D.C. Circuit upheld the Department of Defense designation of Anthropic as a supply-chain security threat on September 25, 2026.
- The ruling rests on 41 U.S.C. § 4713, the broader supply-chain authority in the Federal Acquisition Supply Chain Security Act (FASCSA), not the narrower 10 U.S.C. § 3252 used by the California court.
- The court found DoD acted because Anthropic refused a contract term, not because of the company's public advocacy for ethical AI, distinguishing the First Amendment question from the procurement question.
- The court acknowledged that Claude's built-in restrictions had blocked government users from completing requested tasks on more than one occasion, which it treated as a covered national-security risk.
- A dissent by Circuit Judge Karen LeCraft Henderson argued the majority's broad reading of "manipulate" under FASCSA exceeds congressional intent, which targeted hostile foreign actors, not domestic contractors with ethical policies.
Why it matters
- The ruling creates a direct circuit split: the D.C. Circuit says DoD can blacklist Anthropic under § 4713 without proving bad motive; the Northern District of California says the same action violated the First Amendment under § 3252.
- The Computer and Communications Industry Association warned the decision lets the Pentagon use a tool designed for foreign adversaries against any domestic contractor that declines a contract term.
- TechNet called it a precedent for arbitrary blacklisting without due process, threatening existing business contracts across the defense-tech sector.
- The dissent raised a pointed risk: DoD could now demand Anthropic's replacement agree to remove any AI-use restrictions or face the same fate, effectively forcing model behavior through procurement leverage.
- The case names the defendant "Department of War," signaling an administrative rebranding that itself reflects the militarization context surrounding this dispute.
What to watch next
- Whether Anthropic petitions for en banc review or seeks Supreme Court certiorari to resolve the circuit split, which is the clearest path to a definitive ruling.
- Whether Congress moves to clarify FASCSA's scope, particularly the definition of "manipulate," to close the gap the dissent identified between the statute's foreign-adversary origins and its current domestic application.
- Whether other AI vendors with ethical-use restrictions, such as OpenAI or Google DeepMind, receive similar DoD scrutiny, which would confirm the ruling is being used as a procurement template.
Originally published on Present of AI, a daily source-linked AI news timeline. Read the full timeline or browse the open dataset.