A federal judge in San Francisco called the Trump administration’s ban on Anthropic “troubling” and said it “looks like punishment” during a March 24, 2026 preliminary injunction hearing. U.S. District Judge Rita F. Lin expressed concern that the government was retaliating against Anthropic for publicly opposing Pentagon demands to use its Claude AI for mass surveillance of Americans and fully autonomous weapons. The designation marks the first time a U.S. company has been labeled a “supply chain risk”—a classification typically reserved for foreign adversaries like China or Russia.
Moreover, this isn’t just a corporate dispute. It’s a precedent-setting case that asks whether AI companies can prioritize ethics over government contracts without facing existential threats. Furthermore, the judge’s skepticism suggests courts may check executive overreach when it comes to blacklisting American companies over policy disagreements.
AI Safety vs National Security: The Core Dispute
Anthropic CEO Dario Amodei publicly refused Pentagon demands for “unrestricted military use” of Claude on February 26, 2026, insisting on two red lines: no mass surveillance of Americans, and no fully autonomous weapons without human oversight. The Pentagon demanded Anthropic accept “all lawful purposes” language without the safety restrictions.
“Frontier AI systems are simply not reliable enough to power fully autonomous weapons,” Amodei wrote in his official statement. “Without proper oversight, fully autonomous weapons cannot be relied upon to exercise the critical judgment that highly trained, professional troops exhibit every day.” On mass surveillance, he was equally direct: “Using these systems for mass domestic surveillance is incompatible with democratic values. Consequently, powerful AI makes it possible to assemble scattered, individually innocuous data into a comprehensive picture of any person’s life—automatically and at massive scale.”
The Pentagon refused to provide these assurances. One day later, on February 27, President Trump ordered all federal agencies to immediately stop using Anthropic technology, canceling $200+ million in contracts. Additionally, Defense Secretary Pete Hegseth escalated the personal attacks, calling Amodei “a liar with a God complex.”
This is the heart of the dispute: Anthropic chose AI safety principles over profits. It’s not just employees pushing back—it’s the entire company taking a stand. In fact, this echoes but surpasses the 2018 Google Project Maven employee walkout.
First U.S. Company Labeled National Security Threat
The Pentagon officially designated Anthropic as a “supply chain risk” on March 5, 2026—reportedly the first time this label has been applied to a U.S. company in history. The designation is typically reserved for foreign adversaries. Therefore, it forces defense contractors like Amazon, Microsoft, and Palantir to certify they don’t use Claude in military work.
Legal experts are alarmed. “Supply chain risk designations are typically reserved for foreign adversaries,” notes Just Security’s legal analysis. “Applying this to a U.S. company over a policy disagreement raises serious legal and constitutional questions.” The designation doesn’t just block Anthropic from federal contracts—it threatens the company’s existence by blacklisting it across the entire government-contractor ecosystem.
However, Judge Lin’s comments suggest she sees through the government’s framing. Calling the ban “an attempt to cripple Anthropic” and expressing concern about “retaliating against the company for publicly opposing the Pentagon’s position” indicates the court may view this as punishment masquerading as national security policy.
OpenAI and Musk’s xAI Win Contracts Hours After Ban
The timing raises eyebrows. Hours after Trump banned Anthropic on February 27, OpenAI announced a Pentagon deal worth up to $200 million. Meanwhile, Elon Musk’s xAI—whose founder is a vocal Trump ally—had already secured a contract in January to deploy Grok in classified systems for 3 million military personnel. Both accepted the “all lawful use” terms Anthropic refused.
But OpenAI’s enthusiasm didn’t last. After public backlash, the company amended its contract just four days later on March 2, adding explicit prohibitions on mass domestic surveillance and directing autonomous weapons. CEO Sam Altman later admitted the rollout was “opportunistic and sloppy.” Meanwhile, xAI faced no such pressure—Sen. Elizabeth Warren questioned the Pentagon’s decision to grant classified access to Grok following “the chatbot’s antisemitic posts,” but the contract stands.
The competitive dynamics are clear: AI companies willing to accept unrestricted military use—or at least willing to sign first and add restrictions later—win lucrative government contracts. In contrast, those who draw red lines upfront get banned.
What Happens Next: Judge to Rule “Within Days”
Judge Lin indicated she would rule “within days” on Anthropic’s request for a preliminary injunction to pause the ban. If granted, the ban would be suspended while the full legal case proceeds. This would allow Anthropic to continue operating and defense contractors to resume using Claude. Furthermore, the judge’s comments suggest she’s taking the First Amendment retaliation claims seriously.
The tech community is watching closely. More than 400 Google employees and 75 OpenAI employees signed open letters supporting Anthropic’s AI safety stance. Additionally, another 100+ Google AI workers sent a separate letter to management requesting the same restrictions Anthropic demanded. This echoes the 2018 Project Maven protests, where 4,000+ Google employees forced the company to withdraw from military AI contracts and adopt AI Principles prohibiting weapons and surveillance use.
Sen. Mark Warner (D-VA), Vice Chair of the Senate Intelligence Committee, condemned Trump’s action, saying it “raises serious concerns about whether national security decisions are being driven by careful analysis or political considerations.” The question isn’t just legal—it’s philosophical: Can the U.S. government punish American companies for advocating AI safety principles?
Anthropic’s stand may cost $200+ million in lost contracts, but it’s already won something more valuable: credibility as an AI company willing to defend ethical boundaries. Nevertheless, whether that survives a legal battle against the federal government remains to be seen.
