.png){kind=logo}
A federal judge in California ruled Thursday that the Pentagon unlawfully blacklisted artificial intelligence company Anthropic. U.S. District Judge Rita Lin found that officials violated Anthropic’s First Amendment rights by retaliating over its public criticism.
The August 27 decision followed a dispute over military access to Anthropic’s AI models. The court also blocked directives requiring federal agencies to stop using the company’s technology. Court ruling.
In her 59-page decision, Lin rejected the government’s claim that Anthropic’s public comments justified doubts about its reliability. Officials had argued that the company’s criticism meant they could no longer trust it to protect its models.
“The empty invocation of national security is not a blank check to punish and retaliate against government critics,” Lin wrote.
She accepted that courts should give weight to government judgments on national security. Nevertheless, she found no clear basis for concluding that Anthropic would sabotage its software.
The government’s initial assessment suggested that Anthropic could alter its models after deployment in national security systems. However, Lin said the record showed that the company lacked such access. Officials also conceded that its technology posed no greater risk than other comparable AI models in that respect.
The judge also found that officials denied Anthropic sufficient notice and an opportunity to respond before imposing the restrictions. That conduct violated the company’s rights under the Fifth Amendment.
Lin did not accept every claim. She rejected Anthropic’s separate argument that the president had exceeded his constitutional powers. The government also prevailed on some claims involving agencies that took no relevant action or only temporary measures.
The disagreement began during negotiations over the Pentagon’s use of Claude. Defense Secretary Pete Hegseth sought access to the models for all lawful military purposes.
Anthropic accepted broader military use but sought two limits. It opposed mass surveillance of Americans and fully autonomous weapons that could select and attack targets without human control.
The Pentagon rejected those conditions, arguing that the military needed authority over lawful uses of technology it deployed. Talks became public before the government moved to cut ties with the company.
In February, President Donald Trump directed federal agencies to stop using Anthropic’s products. Hegseth also announced the supply chain risk action, which the department formalized in March.
The designation barred defense contractors from using Anthropic’s technology in work for the department. Anthropic then challenged the restrictions through lawsuits in California and Washington, D.C.
Lin had already issued temporary relief in March, pausing enforcement while the California lawsuit proceeded. Her latest decision went beyond that earlier order and found the challenged government actions unlawful.
However, the Pentagon relied on two separate legal authorities for its supply chain risk designations. That required Anthropic to pursue challenges in different courts.
The Washington case continues, leaving a separate designation in place despite the California ruling. As a result, Thursday’s decision does not fully resolve Anthropic’s legal position with the federal government. Case status.
“We welcome the court’s ruling that this supply chain risk designation was unlawful,” an Anthropic spokesperson said.
The company said it wants to restore the business conditions that existed before the blacklisting. It also expressed interest in working with the government on national security.
The ruling does not require the Pentagon to renew contracts or resume using Claude. The department can move to other AI providers through lawful purchasing decisions.
ALSO READ: US Weighs New Chip Tariffs, AI Data Centers Face Risk
