A federal judge has sided with Anthropic in its fight with the Pentagon, ruling that the U.S. government unlawfully retaliated against the AI safety company after it refused to allow its Claude model to be used for fully autonomous weapons and domestic surveillance.
The ruling strikes down the Pentagon’s decision to designate Anthropic a national security supply chain risk and finds that the government violated the company’s First and Fifth Amendment rights. It also gives Anthropic a major legal victory in a dispute that has produced growing tensions over how far AI companies should be expected to go when building technology for the military.
This ruling came after months of tension between Anthropic and the Pentagon over the company’s refusal on how Claude could be used by the U.S. military.
How The Dispute Started
The disagreement centered on two uses of Claude that Anthropic refused to permit. The company opposed mass domestic surveillance of Americans and fully autonomous weapons.
Anthropic said it remained willing to support lawful national security uses of its technology but argued that current AI models were not reliable enough for fully autonomous weapons. It also raised concerns about using AI for domestic surveillance.
The negotiations eventually broke down. In February, Defense Secretary Pete Hegseth directed the government to designate Anthropic as a national security supply chain risk. President Donald Trump separately ordered federal agencies to stop using Anthropic’s technology, while the Pentagon moved to prevent defense contractors from doing business with the company.
Anthropic sued in March, arguing that the government’s actions were retaliatory and that the supply chain designation had no legal or factual basis.
Judge Rejects Pentagon’s Reasoning
U.S. District Judge Rita F. Lin found that the government’s own record did not support its explanation for the blacklist.
The court noted that the government had provided little evidence showing that Anthropic posed an actual risk of sabotaging or compromising military systems. The judge also pointed to the government’s continued interest in working with Anthropic even after the company had been designated a supply chain risk.
Ultimately, the ruling found that the designation violated the statutory rules governing supply chain risks and was “arbitrary and capricious.” Judge Lin also rejected the government’s reliance on national security as sufficient justification for its actions.
“The empty invocation of national security is not a blank check to punish and retaliate against government critics,” Lin said.
The court granted Anthropic summary judgment on its First Amendment and due process claims and ordered permanent injunctive relief. The ruling also set aside the Pentagon’s supply chain designation.
The Fight Is Still Not Over
Anthropic’s court victory has not completely settled its relationship with the U.S. government.
On September 3, Pentagon official Emil Michael said Anthropic was still designated a supply chain risk for the Defense Industrial Base, despite the court ruling. The statement came shortly after Commerce Secretary Howard Lutnick said the administration trusted Anthropic and that the company was back in good standing with the Trump administration.
The legal fight itself is also not finished. The Department of Justice is expected to appeal Judge Lin’s ruling, and a separate Anthropic lawsuit challenging a related Pentagon designation is still being considered by a federal court in Washington. This means the question of how far the government can go in punishing an AI company over its safety refusals may not be fully settled until both cases run their course.
For Anthropic, the ruling still represents a significant legal victory, as the company successfully challenged the government’s attempt to punish it over its refusal to use its model for surveillance.
The case also establishes an important boundary for the government’s growing reliance on AI. National security concerns remain a major consideration in military technology decisions, but Judge Lin’s ruling makes clear that those concerns still have to be supported by law and evidence.
