(SeaPRwire) – By: Julian Holbrooke
The Pentagon blacklisted an AI company for refusing to deploy its models on autonomous weapons. A federal judge just handed the company a 59-page ruling that reads like a textbook case of executive overreach. Judge Rita Lin called the designation “arbitrary and capricious.” She said Secretary of War Pete Hegseth “overstepped his authority.” She ruled the action was unlawful retaliation under the First Amendment. And she said Anthropic was denied due process under the Fifth. The empty invocation of national security, Lin wrote, “is not a blank check to punish and retaliate against government critics.”
The official Washington line was straightforward. Anthropic refused to drop safeguards on military AI use. Hegseth designated the company a “supply-chain risk to national security.” President Trump ordered federal agencies to cease using Anthropic technology on February 27, branding it a “radical Left AI company.” Hegseth declared Pentagon partners could no longer conduct commercial activity with Anthropic. When Anthropic pushed back, he compared the company’s red lines to “Boeing giving us airplanes and telling us who we can shoot at.” He called CEO Dario Amodei an “ideological lunatic.” The language was deliberately provocative. The intent was deterrence. Make an example of one company, and the rest will comply. The Boeing analogy was not accidental. It framed Anthropic’s safety conditions as interference with military operations. The message to every other AI lab was clear. Bend, or get blacklisted.
The real story sits underneath that language. Anthropic had a Pentagon agreement worth up to $200 million. Claude was deployed for various military operations. The company was deeply embedded. Amodei’s position was not ideological extremism. He supported military use of Claude. He simply argued it should not be used for mass domestic surveillance or guiding fully autonomous weapons. Current frontier models are not reliable enough for the latter task. The Pentagon wanted “any lawful use” access with no conditions. The defense establishment knew Anthropic was the better model. One defense official told Axios in February, “the problem for these guys is they are that good.” Lin acknowledged the Pentagon “is undisputedly free to select the AI vendor of its choice.” But she drew the line. The government can fire a vendor. It cannot weaponize the designation process to punish protected speech.
The pendulum has shifted. The supply-chain-risk designation is no longer a free tool for executive retaliation. Any AI company drawing red lines now has constitutional leverage. The designation is subject to judicial review. The “arbitrary and capricious” standard is not a formality. Anthropic has another lawsuit pending in Washington, DC over a supply-chain designation that could bar it from civilian government contracts. That one will test the same playbook against a different agency. The Pentagon can still shop for other vendors. But the precedent is set. Naming a company a national security threat because it publicly criticizes the government is retaliation, not regulation.
Author bio: Julian Holbrooke, an overseas international relations analyst who frequently contributes to major European daily newspapers.