A U.S. judge recently put a stop to the Pentagon’s decision to blacklist Anthropic, which is currently embroiled in a significant conflict with the military regarding AI safety in combat scenarios.
In a lawsuit filed in federal court in California, Anthropic contends that Defense Secretary Pete Hegseth exceeded his authority by classifying the company as a national security supply-chain threat. This classification allows the government to label businesses that might pose risks to military systems due to potential infiltration or sabotage by enemy forces.
This unprecedented action by Hegseth prevented Anthropic from securing certain military contracts. The move came after Anthropic declined to permit the military to utilize its Claude AI models for U.S. surveillance or autonomous weaponry.
Executives at Anthropic have indicated that this could lead to billions in losses, both financially and in terms of reputation.
U.S. District Judge Rita Lin, appointed by former President Joe Biden, issued a 59-page ruling, deeming the Pentagon’s decision “illegal and baseless.”
In her order, she stated, “The empty invocation of national security is not a blank check to punish and retaliate against government critics.”
Anthropic embraced the ruling, expressing in a statement its commitment to collaborate productively with the government to leverage AI for national security in a way that benefits all Americans.
The Pentagon has not yet provided a response regarding the ruling.
While Anthropic argues that AI models aren’t dependable enough for use in autonomous weapons and that it opposes domestic surveillance as a rights infringement, the Pentagon asserts that private firms shouldn’t be allowed to limit military capabilities.
This designation marks the first instance in which a U.S. company has been publicly classified as a supply-chain risk under a rarely used government-procurement regulation intended to safeguard military systems from foreign sabotage.
In its lawsuit filed on March 9, Anthropic claimed that the government violated its First Amendment rights by responding negatively to its views on AI safety and further stated that it was denied the opportunity to contest the designation, infringing on its Fifth Amendment right to due process.
The lawsuit claims the decision lacked legal grounds, was not factually supported, and contradicted the military’s previous commendations of Claude.
On the other hand, the Justice Department argued that Anthropic’s unwillingness to lift the restrictions might create uncertainty in the Pentagon regarding the use of Claude and could potentially disrupt military systems during operations, as stated in a court filing. The government also insisted that the classification originated from Anthropic’s refusal to comply with contractual terms, rather than its stances on AI safety.
Additionally, Anthropic has another lawsuit currently pending in Washington, D.C., concerning a different Pentagon supply-chain risk designation that might result in its exclusion from civilian government contracts.
