- The Pentagon invoked a law designed for foreign sabotage threats against a domestic U.S. company, something that has never been done before.
- Defense Secretary Hegseth praised Claude as “exquisite” technology just days before declaring its maker a national security risk.
- Anthropic’s First Amendment and due process claims are strengthened by Trump and Hegseth’s own public statements attacking the company.
- The U.S. military was actively using Claude in operations even as the Pentagon moved to ban Anthropic from government contracts.
The clash between Anthropic and the Department of Defense has escalated into one of the most consequential legal fights in recent tech history, and the early read from national security law specialists is that the Trump administration may have seriously miscalculated.
The Pentagon designated Anthropic a national security supply chain risk on March 3, effectively blocking the AI company from federal contracts after it refused to remove restrictions on its Claude AI model that prohibit the military from using it for autonomous weapons systems or domestic surveillance.
Anthropic fired back with a lawsuit on Monday, arguing the designation violates its constitutional rights and has no legitimate legal basis. Legal experts who reviewed the case largely agree.
An Obscure Law Used in an Unprecedented Way
At the heart of the dispute is Section 3252, a rarely invoked statute that allows the Defense Secretary to exclude companies from certain contracts if they pose a risk of enemy infiltration or sabotage of military information systems. The catch is that the law was designed with foreign adversaries in mind, not American companies operating entirely within U.S. borders.
Anthropic is incorporated and headquartered in the United States. It has no foreign entanglement that legal experts can identify. University of Minnesota Law School professor Alan Rozenshtein put it plainly: it is not at all clear the statute can even apply to a domestic company in this situation.
A Reuters review of legal databases found no prior instance of the law being used against a U.S. company, and no public record of any company ever being designated a supply chain risk under it.
Amos Toh of the Brennan Center for Justice was equally direct. Anthropic’s usage policies are safety protocols, he noted, and whatever one thinks of their merit, they bear no resemblance to the foreign sabotage threats the law was written to address.
The Contradictions in the Government’s Position
What may hurt the Pentagon’s case most is not just the legal technicalities but the sheer internal inconsistency of its own conduct. Defense Secretary Pete Hegseth reportedly praised Claude as “exquisite” technology the Defense Department would “love” to work with during a February meeting with Anthropic executives.
The U.S. military used Claude as recently as last month during strikes on Iran. And yet, just days later, Hegseth declared the same tool a national security threat and moved to ban the company that makes it from government contracts.
Rozenshtein described the contradiction bluntly: the government was simultaneously threatening to force Anthropic to sell its services, actively using those services in military operations, and declaring them too dangerous for government contracts. Those positions simply cannot all be true at once.
Joel Dodge of Vanderbilt University pointed to the public rhetoric from both Hegseth and President Trump, including a post in which Trump called Anthropic a “RADICAL LEFT WOKE COMPANY,” as evidence of personal animus rather than legitimate national security reasoning.
That kind of paper trail, Dodge argued, gives Anthropic a solid foundation for its First Amendment claim that the designation was meant to punish the company for its views on AI safety.
Due Process and the “Death Penalty” Problem
Anthropic’s lawsuit also takes aim at how the designation was carried out, arguing it imposed severe penalties with no factual findings, no meaningful explanation and no opportunity for the company to respond or challenge the decision. That is a textbook due process claim under the Fifth Amendment, and legal experts say it has real teeth.
Government contract specialist Eric Crusius described the designation as the corporate equivalent of the death penalty, and noted the Pentagon would need to demonstrate it had no alternative and had exhaustively considered other options before taking action this drastic.
The potential financial consequences are enormous. Anthropic executives have said the designation could cost the company multiple billions of dollars in 2026 revenue alone, to say nothing of the reputational damage to a company that has spent years positioning itself as the responsible actor in the AI industry.
Courts will almost certainly defer to executive branch judgment on matters of genuine national security. That deference, though, has limits, and it tends to weaken considerably when the government’s own actions suggest the decision was driven more by political grievance than by any coherent security rationale.
Follow TechBSB For More Updates
