Top News

In the annals of unusual corporate disputes, few will match what is currently unfolding between Anthropic — the San Francisco AI company behind Claude — and the United States Department of Defense. The Pentagon has effectively blacklisted one of America’s most valuable AI companies, using a national security law previously deployed exactly once in history, because Anthropic refused to remove its ethical guardrails on autonomous weapons and mass surveillance.

Anthropic filed two lawsuits on March 9 to fight back. The outcome could reshape the relationship between Silicon Valley and Washington for a generation.

How It Started: Two Red Lines the Pentagon Couldn’t Accept

The dispute traces back to a negotiation that broke down in early 2026. The Pentagon wanted access to Claude — Anthropic’s AI model — for defense applications. Anthropic agreed, with conditions. Two of them proved fatal to the deal.

Anthropic said Claude would not be used for mass surveillance of American citizens. And it would not be deployed in fully autonomous lethal weapons systems — situations where an AI makes targeting and firing decisions without meaningful human oversight.

For Anthropic, these were non-negotiable ethical commitments baked into the company’s founding mission. For the Pentagon, operating in an environment of active conflict including ongoing operations to the Iran war, they were unacceptable constraints on national security capability.

Negotiations collapsed. The Trump administration directed federal agencies to stop using Anthropic’s technology. And then the Pentagon reached for a legal instrument that had never been used against an American company before.

The Weapon: A Law Built for Chinese Spyware

The Federal Acquisition Supply Chain Security Act — FASCSA — was designed to protect U.S. government systems from foreign technology threats. Think Huawei. Think Chinese-linked software embedded in critical infrastructure. The kind of supply chain risk that comes from adversarial foreign actors, not domestic AI labs in San Francisco.

Before Anthropic, FASCSA had been used exactly once — against Acronis AG, a Swiss cybersecurity firm, in September 2025.

On March 4–5, 2026, the Pentagon used it against Anthropic. The designation formally labels Anthropic and its products a “supply chain risk” — a classification that bars government contractors from using Anthropic technology in DoD- work, effectively cutting the company off from the entire defense ecosystem and large parts of the federal contracting market.

The practical consequence: over $200 million in Pentagon- contracts canceled or threatened immediately, with billions more in broader enterprise revenue at risk as companies with any DoD exposure began auditing their Anthropic usage and demanding compliance certifications.

The Lawsuits: Two Courts, Two Arguments

Anthropic’s legal response came on March 9 — two lawsuits, two courts, two distinct legal theories.

The first was filed in the U.S. District Court for the Northern District of California, challenging the DoD’s designation under 10 USC § 3252, which governs armed services procurement. This lawsuit seeks a temporary restraining order to halt enforcement while the broader case is heard.

The second — the one generating the most legal attention — was filed directly in the U.S. Court of Appeals for the D.C. Circuit. Under FASCSA, challenges to supply chain risk designations must be brought in the DC Circuit. This petition asks the court to review and overturn the Pentagon’s designation, halt its enforcement, and declare it an unlawful abuse of statutory authority.

Anthropic’s core argument: the FASCSA designation is not a genuine supply chain security determination. It is ideological retaliation — the government punishing a company for its publicly stated position on AI safety. FASCSA, Anthropic argues, requires the “least restrictive means” and exists to protect the government from threats, not to punish American suppliers for refusing to comply with demands the company considers unethical. Using it this way, Anthropic contends, also violates the First Amendment — penalising protected speech — and the Fifth Amendment’s due process guarantees.

The Stakes: Existential for Anthropic, Precedent-Setting for Everyone Else

For Anthropic, valued at over $380 billion at its peak, this is not a routine regulatory dispute. The supply chain designation threatens the company’s ability to operate in government and enterprise markets, spooks investors, and potentially forces partners including Amazon — a major Anthropic backer — to navigate their own compliance exposure.

But the stakes extend well beyond one company.

Every major AI lab in America is watching this case. If the government can blacklist a U.S. AI firm for refusing to drop safety limits on autonomous weapons — using a law designed for foreign adversaries — the precedent is chilling. OpenAI, Google DeepMind, Meta AI, and others all maintain their own versions of ethical use policies. The Anthropic case tests whether those policies can survive contact with a government that disagrees with them.

OpenAI reportedly moved quickly to fill the contracts Anthropic lost. Whether that is opportunism or a warning about what compliance with Pentagon demands actually looks like in practice is a question the industry is now forced to confront openly.

What Contractors Need to Do Right Now

For the thousands of companies that use Anthropic’s technology and hold government contracts, the immediate question is practical. The key threshold: does your contract incorporate FAR 52.204-30 — the provision implementing FASCSA? If it does, and if Anthropic products are used in performance of that contract, you have a potential compliance issue.

Legal experts recommend beginning a reasonable mapping of Anthropic usage across your systems immediately — both to assess actual exposure and to demonstrate good-faith compliance preparation. However, given the legal uncertainty and Anthropic’s pending motion for a stay, most advisors suggest holding off on immediately notifying the government under FAR 52.204-30(c)’s three-business-day requirement, and instead monitoring how the courts respond in the coming weeks.

If Anthropic wins a stay — pausing enforcement of the designation while the cases are heard — the compliance calculus changes significantly.

The Bigger Question

At its heart, this dispute is about a question that has no clean legal answer yet: can a private company set ethical limits on how a government uses its technology, and can the government retaliate when it doesn’t like those limits?

Anthropic’s position is that it can and must. The Pentagon’s position is that national security needs do not yield to the ethical preferences of private suppliers. The courts will now decide which principle prevails.

The cases are in their earliest stages. No rulings have been issued. Hearings could be weeks or months away. But the outcome — whatever it is — will define the terms on which American AI companies engage with their own government for years to come.

An AI company built to make artificial intelligence safe for humanity is now fighting the U.S. government in two federal courts to preserve that mission. Whatever you think of the merits, it is one of the more consequential corporate legal battles of the decade — and it is only just beginning.

Cases filed: March 9, 2026. U.S. District Court, Northern District of California + U.S. Court of Appeals, D.C. Circuit. Status: Early stage, no rulings as of March 12, 2026.