Judge tears up Pentagon's Anthropic blacklist as retaliation

Judge Rita Lin didn't hedge. The Pentagon's designation of Anthropic as a supply-chain risk — an order from Defense Secretary Pete Hegseth telling every federal agency, defense or not, to stop contracting with the company — was unlawful retaliation under the First Amendment, "arbitrary and capricious," and a Fifth Amendment due-process failure. Her line: "The empty invocation of national security is not a blank check to punish and retaliate against government critics."

The trigger was Anthropic refusing to strip guardrails that keep Claude out of fully autonomous weapons and mass surveillance of Americans. Lin found the record showed the government wanted to make a public example of Anthropic for its "arrogance" in criticizing it. Anthropic filed two suits back in March — California (this one) and Washington, D.C., which is still live, so this is a first win, not the last word.

Elsewhere in the same 48 hours: Hugging Face's Pollen Robotics opened pre-orders for Microduck, a $399, 25cm, sub-800g biped with 15 motors, a camera and LiDAR — training environments, reward functions and the sim-to-real recipe all public on GitHub. And 100+ AI labs and security firms signed an open letter pushing coordinated defense against AI-driven attacks on hospitals and infrastructure.

Why it matters: a US court just held that an AI company's safety policy is protected speech the government can't punish it for.