A federal judge in San Francisco has temporarily blocked the Trump administration’s attempt to ban the AI startup Anthropic from government work, calling the administration’s actions a likely case of illegal retaliation.

The Ruling

U.S. District Judge Rita Lin issued a preliminary injunction on Thursday, pausing the Pentagon’s designation of Anthropic as a “supply chain risk.” The judge also froze a directive from President Trump that ordered all federal agencies to stop using the company’s technology.

In a sharp 43-page ruling, Judge Lin described the government’s move as “arbitrary and capricious,” suggesting it was designed to punish the company for its public stance on AI safety rather than to protect national security. She noted that while the government is free to stop using a specific vendor, it cannot use “Orwellian” labels to brand an American company a “saboteur” simply for disagreeing with policy.

The Core Dispute

The legal battle began after negotiations between Anthropic and the Department of Defense (referred to by the administration as the “Department of War”) collapsed. The conflict centered on how Anthropic’s AI model, Claude, could be used:

• Anthropic’s Position: The company insisted on safeguards to prevent its technology from being used for mass domestic surveillance or the development of fully autonomous lethal weapons.
• The Administration’s Position: Officials demanded “unfettered access” for all “lawful uses.” When Anthropic refused to waive its safety guardrails, Defense Secretary Pete Hegseth labeled the company a national security risk—a designation usually reserved for foreign adversaries like Huawei.

Impact and Background

The ruling is a significant victory for Anthropic, which argued that the “supply chain risk” tag would cause irreparable reputational damage and billions in lost revenue. The company’s technology is currently deeply embedded in various government systems, including classified military operations.

President Trump had previously attacked Anthropic on social media, calling it a “Radical Left” company. Following the breakdown in talks, the administration had already begun a six-month phase-out of Claude in favor of models from competitors like OpenAI.

What’s Next

The injunction will take effect in seven days, giving the Trump administration time to file an appeal. While the ruling pauses the immediate blacklisting, the broader legal battle over the administration’s authority to sanction domestic tech companies is expected to continue for months. Anthropic still has a parallel lawsuit pending in Washington, D.C., addressing separate aspects of the federal ban.