Judge Blocks Pentagon Blacklist of Anthropic as Unlawful

A federal judge has permanently blocked the Pentagon from enforcing its designation of artificial intelligence company Anthropic as a national security supply chain risk. District Judge Rita Lin issued the 59-page ruling late Thursday, finding that the Department of Defense acted unlawfully by punishing the firm for its public criticism of military plans to deploy AI in autonomous weapons and mass surveillance programs.

The decision centers on actions taken earlier this year under the Trump administration. After Anthropic refused to remove safeguards from its Claude models that would permit their use in fully autonomous lethal systems or widespread domestic monitoring, Defense Secretary Pete Hegseth declared the company a supply chain risk. That label effectively barred other defense contractors from working with Anthropic, cutting the firm off from significant government-related business. Lin ruled that these steps amounted to unconstitutional retaliation against protected speech rather than a legitimate security measure.

In her order, Lin stated that the empty invocation of national security provides no blank check to punish government critics. She found the designation stemmed primarily from a desire to make an example of Anthropic for what officials viewed as arrogance in questioning administration policy. The judge also concluded that the moves violated the company's due process rights by imposing sweeping penalties without adequate basis in the cited federal statutes.

Anthropic had sought to limit its technology to lawful applications while expressing concerns over the reliability of current AI systems for battlefield decisions without human oversight. The company maintained that its objections reflected technical limitations rather than opposition to national defense. Following the ruling, an Anthropic spokesperson welcomed the outcome and expressed continued interest in productive collaboration with government agencies on AI applications that benefit national security.

The Pentagon had argued that unrestricted access to advanced models posed risks if safeguards prevented full operational use. Administration officials framed the supply chain designation as necessary to protect classified systems and maintain contractor reliability. Lin rejected those justifications as insufficient to override First Amendment protections, noting that the government retains options to select other AI providers without resorting to punitive blacklisting.

This case highlights recurring tensions between executive branch assertions of broad authority in security matters and judicial scrutiny of potential overreach. Past administrations of both parties have invoked similar national security rationales to sideline private entities, often with mixed results in court. The ruling does not compel the military to adopt Anthropic's models or prevent it from awarding contracts elsewhere, provided actions comply with regulations and avoid viewpoint discrimination.

The government is expected to appeal the decision, which could eventually reach higher courts for clarification on the limits of supply chain risk designations. Such reviews typically examine whether agencies followed statutory procedures or crossed into content-based punishment. For now, the order restores Anthropic's ability to engage with federal contractors without the prior restrictions.

Observers note that the dispute arose amid rapid advances in AI capabilities and debates over appropriate guardrails. Companies developing these tools often face pressure to balance commercial interests, ethical concerns, and government demands. Lin's emphasis on evidence-based decision making rather than reflexive security claims aligns with longstanding skepticism toward expansive bureaucratic power that can stifle innovation and dissent. The outcome may encourage other firms to voice reservations about specific applications without fear of administrative retaliation.