Defense Secretary Pete Hegseth wanted AI company Anthropic to give the Pentagon broad access to its technology. Anthropic said no.
The company refused to allow its AI models to be used for mass surveillance of Americans or the development of deadly autonomous weapons.
So the Trump administration decided to punish the company by designating Anthropic a “supply chain risk.”
Now, a federal judge has ruled that the Pentagon’s retaliation was illegal, and accused the government of trying to “make a public example” out of the company for criticizing it.
That’s a pretty brutal finding for Hegseth’s Defense Department.
U.S. District Judge Rita Lin ruled Thursday that the government’s decision to designate Anthropic a “supply chain risk” violated the company’s First Amendment and due process rights.
And she wasn’t impressed by the administration’s repeated invocation of national security.
“The empty invocation of national security is not a blank check to punish and retaliate against government critics,” Lin wrote.
The case began after Anthropic refused to give the Pentagon unrestricted use of its AI models.
The company had already been doing serious business with the military. It previously held a $200 million Pentagon contract, and as of February, its Claude large language model was reportedly the only AI model available in the military’s classified systems, according to Axios.
But Anthropic established limits on how its technology could be used.
Among them: no mass surveillance of Americans and no use of its models to develop autonomous weapons capable of making lethal decisions.
The Trump administration didn’t take kindly to those restrictions.
Hegseth reportedly met with Anthropic CEO Dario Amodei and issued the company an ultimatum.
When Anthropic refused to back down, the Pentagon escalated.
The administration declared Anthropic a “supply chain risk” — a designation that had previously been reserved for foreign adversaries.
The consequences were significant.
The Pentagon was barred from using Anthropic’s models, while military contractors were required to certify that they weren’t using Anthropic tools in their own workflows.
Anthropic responded with two lawsuits, arguing that the designation was unprecedented and unlawful.
The company eventually got its day in court. And the judge wasn’t buying the government’s argument.
Lin found that the administration’s decision violated Anthropic’s constitutional rights. More importantly, she concluded that the government’s actions weren’t simply about protecting national security.
According to Lin, the decision was “based on a desire to make a public example out of Anthropic for its ‘arrogance’ in criticizing the government.”
That’s a very different explanation.
The Pentagon said it was protecting national security.
The judge said the government was retaliating against a critic.
Then there was another problem for the administration.
Other government departments continued using Anthropic’s technology even after the company was declared a supply chain risk.
Lin said that undermined the government’s claim that Anthropic posed a serious security threat.
The administration had argued that it was concerned Anthropic could potentially “poison its software to harm national security.”
Yet other government agencies apparently didn’t find the company’s technology too dangerous to keep using.
That contradiction did not go unnoticed.
Lin had already issued a preliminary order blocking the designation, saying the government’s action was “likely both contrary to law and arbitrary and capricious.”
Thursday’s ruling goes further.
The case also puts a spotlight on a larger fight that is rapidly developing inside the technology industry.
AI companies are increasingly becoming contractors for the federal government and the military, but some are drawing lines around how their technology can be used.
Anthropic drew one of those lines.
And instead of simply accepting the company’s decision, the Pentagon tried to punish it.
President Donald Trump then personally piled on.
In a social media tirade, Trump called Anthropic a “RADICAL LEFT, WOKE COMPANY” and accused it of trying to “DICTATE HOW OUR GREAT MILITARY FIGHTS AND WINS WARS!”
Anthropic CEO Dario Amodei offered a very different explanation for the administration’s anger.
In a memo to employees, Amodei suggested that Anthropic had become a target because it hadn’t donated millions of dollars or offered “dictator-style praise to Trump” like competitor OpenAI.
And then came a remarkable bit of timing.
Just hours after Anthropic was declared a supply chain risk, OpenAI struck a deal allowing the Pentagon to use its AI tools in classified military systems.
That doesn’t prove that politics played a role in the administration’s decisions.
But it certainly makes the sequence of events worth examining.
Anthropic has said it remains willing to work with the government and wants AI technology to contribute to national security while protecting the public.
The company welcomed Lin’s ruling and said it remains focused on working productively with the government.
But the court’s message to the administration was considerably less diplomatic.
The government can protect national security. It can set rules for military contractors. It can decide what technology it wants to use.
What it apparently cannot do is invoke national security as a magic phrase whenever a company refuses to comply with its demands, and then retaliate against that company for speaking out.
