AI Giant’s Court Battle: National Security or Retaliation?

A gavel held above a sounding block with a person reading documents in the background

The Pentagon’s decision to slap a “supply chain risk” label on a U.S. AI company has triggered a courtroom fight that tests how far the federal government can go before national-security power starts looking like political punishment.

Quick Take

Anthropic sued after the Pentagon designated its Claude AI a “supply chain risk,” cutting it off from military-related work.
Anthropic argues the designation functions as retaliation for its public push to restrict AI use in mass surveillance and fully autonomous weapons.
The Trump administration directed agencies to phase out Claude, while the Pentagon also announced a six-month wind-down to limit disruption.
Legal analysts question whether the “supply chain risk” tool—typically used for foreign threats—was applied using the least restrictive means required by law.

Why the “Supply Chain Risk” Label Matters in a Wartime AI Environment

The dispute started in early March 2026 after the Pentagon designated Anthropic as a “supply chain risk,” a label that effectively blocks its Claude system from Defense Department contracting and related military use. Claude had been used for intelligence processing and targeting support connected to the U.S. war with Iran, raising the stakes for continuity and reliability. After the designation, contractors and agencies began removing Claude and certifying non-use in Defense work.

Anthropic responded by filing two lawsuits on March 9, 2026—one in U.S. District Court for the Northern District of California and another in the D.C. Circuit. The company says the government crossed a constitutional line by using a procurement-based national security mechanism as leverage against speech and policy advocacy. The Pentagon has largely declined to comment publicly on the litigation, emphasizing operational control concerns instead.

Contract Terms at the Center: “All Lawful Use” Versus Built-In Restrictions

At the center of the clash are contract terms, not just politics. The negotiations where the Pentagon wanted language allowing “all lawful use,” while Anthropic insisted on safeguards that would restrict Claude from being used for mass domestic surveillance or for fully autonomous weapons. Those safeguards match the company’s public posture on AI safety and limits. Negotiations reportedly failed when Anthropic refused to remove those restrictions.

The Trump administration’s posture was not subtle. President Trump publicly framed Anthropic as ideologically “Radical Left,” and the administration moved to phase out Claude across federal use connected to Defense work. The Pentagon, under Secretary Pete Hegseth— leading a rebranded “Department of War”—also announced an “immediate stop” paired with a six-month phase-out designed to reduce disruption, a timeline that later became part of the legal debate.

Legal Questions: National Security Tool or Viewpoint Retaliation?

Anthropic’s lawsuits argue the “supply chain risk” label was applied in a way that violates First Amendment protections, with the company claiming the government punished it for advocating limits on surveillance and autonomous weapons. The government’s position is that this is about command authority and battlefield decision-making: vendors should not dictate operational constraints during an active conflict, and procurement choices belong to the military.

Even some analysts sympathetic to strong national defense questioned the mechanics. Commentary highlighted that procurement law in this area is supposed to use the least restrictive means to address risk, and critics argued the record described does not show the kind of urgent, evidence-backed threat normally associated with a sweeping exclusion. The existence of a six-month wind-down has also been cited as complicating claims of immediate, catastrophic risk.

Ripple Effects: Contractors, Competitors, and a Precedent for U.S. Firms

The practical impact is broader than one vendor. Defense contractors were required to certify they were not using Claude on Defense work, and Anthropic faced the prospect of losing “hundreds of millions” tied to government-related revenue. Meanwhile, OpenAI—Anthropic’s major competitor—struck an alternative arrangement with the Pentagon after Anthropic’s talks collapsed. Details of that deal were criticized in commentary as potentially “sloppy,” and later suggested it faced renegotiation pressure.

The long-term question is whether the “supply chain risk” label becomes a precedent for domestic companies whose internal policies, safety commitments, or public speech collide with Washington’s short-term demands. Conservatives should care about both sides of that tension: a military fighting a real war needs tools it can rely on, but constitutional guardrails exist precisely to prevent the government from using national-security bureaucracy as a workaround to punish disfavored viewpoints. The courts will decide what the facts support—and what limits still hold.

Pentagon responds to Anthropic suit arguing the company is a 'substantial risk' to national security https://t.co/JFozP0mHMx

In a new filing, the Department of Defense outlined its arguments against Anthropic's lawsuit.
Dado Ruvic/REUTERS

In a new filing, the Department of D…

— America's Pick (@nims213) March 18, 2026

For now, the most defensible conclusion is that this case sits at an uncomfortable intersection: wartime urgency, fast-moving AI capabilities, and federal procurement power that can crush a company overnight. If the government can’t show a clear, procurement-relevant risk supported by evidence and tailored restrictions, the “risk” label starts to look less like security and more like a tool future administrations could turn on any domestic target—left or right.