Anthropic takes the Pentagon to court over AI blacklisting

Anthropic just took the gloves off and sued the Pentagon.

On Monday, the AI lab filed dual lawsuits against the Department of Defense and the Trump administration in California and Washington, D.C. The legal showdown follows the Pentagon slapping Anthropic with a "supply-chain risk" designation, an aggressive blacklisting tool usually reserved for foreign adversaries like Huawei.

Two red lines that triggered a blacklist

The dispute centers on acceptable military use cases for frontier AI. Anthropic drew two firm red lines: it refused to allow Claude to power mass domestic surveillance of Americans, and it barred the tech from operating fully autonomous weapons where algorithms pull the trigger without a human in the loop.

Defense Secretary Pete Hegseth pushed back, arguing the military should be free to deploy advanced AI for "any lawful purpose" without arbitrary limits set by a private contractor. Hegseth and President Donald Trump publicly branded Anthropic and CEO Dario Amodei as "woke" and "radical."

Retaliation came fast. Trump ordered every federal agency to purge Anthropic tech within six months. The General Services Administration tore up Anthropic's OneGov contract, effectively cutting off Claude across all three branches of the federal government. Agencies from the Treasury to the State Department quickly scrambled to drop the company.

A dual-court counterattack

Anthropic isn't going down without a fight. In its San Francisco complaint, the company argues the administration illegally weaponized executive power to punish protected speech under the First Amendment and violated Fifth Amendment due process.

Anthropic contends its warnings about model limitations and refusal to enable autonomous lethal warfare are protected viewpoints. The government can choose not to buy its software, but it cannot legally use state power to destroy the business for speaking up.

In the D.C. Circuit Court of Appeals, Anthropic hit the Pentagon on federal procurement rules. By law, excluding a vendor requires a formal risk assessment, written notice, an opportunity to respond, and congressional notification. The Pentagon skipped all of it.

Meanwhile, enterprise partners like Microsoft are caught in the crossfire. Microsoft says it still works with Anthropic, but it is now forced to build operational firewalls to keep Anthropic tech far away from its lucrative Pentagon contracts.

Why it matters

This is arguably the most consequential clash between Silicon Valley and the national security state in modern history.

For enterprise builders and developers, the fallout is immediate. A supply-chain risk label does not just hit Anthropic; it threatens any contractor or vendor touching the Pentagon. Companies building on top of Claude now face potential cross-contamination risks, underscoring why vendor lock-in with a single frontier model is dangerous.

For the AI industry, this case will define who ultimately governs artificial intelligence. If the Pentagon can crush a private lab for setting safety guardrails on lethal drones and mass spying, every rival AI firm will be forced to roll over and hand over unchecked access to retain government favor.

We think Anthropic made the only move it realistically could. Backing down would have torched its entire identity and market positioning as the responsible AI standard. By challenging the DOD on both First Amendment retaliation and administrative procedure, Anthropic has a very strong legal shot at freezing an unprecedented executive overreach.