A federal appeals court has delivered a significant victory to the Pentagon in its ongoing conflict with a prominent artificial intelligence firm, a decision that could redefine oversight of military AI in the United States.
In a 2-1 vote, a three-judge panel backed the Defense Department, affirming its choice to completely exclude Anthropic from defense contracts.
The Fight That Split Two Federal Courts
The controversy began in February when Defense Secretary Pete Hegseth requested that Anthropic remove safety barriers that prevent Claude from supporting fully autonomous lethal weapons and large-scale domestic surveillance. Anthropic publicly declined on February 26, shortly before Hegseth’s deadline passed.
On March 3, Hegseth officially labeled Anthropic a national security risk. This classification terminated the company’s active military agreements and prohibited other defense contractors from utilizing its technology.
According to the court documentation, an executive at Anthropic had previously raised questions regarding a contractor’s implementation of Claude during a specific military operation.
Case records note that media coverage identified this mission as the January 3 operation that resulted in the capture of Venezuelan President Nicolás Maduro.
Defense representatives pointed to that event as proof that Claude’s performance remained too unpredictable for defense applications.
The majority opinion was written by Judge Gregory Katsas alongside Judge Neomi Rao, who stated that the Pentagon possessed “ample support” for its determination. Judge Karen LeCraft Henderson issued a dissenting opinion.
Why Anthropic Says This Fight Isn’t Over
Just last month, a separate federal court in California reached a conflicting decision. Judge Rita Lin determined that a broader, related ban by the Trump administration was unconstitutional, describing it as punitive action intended to make “a public example” out of Anthropic.
Anthropic issued a strong response to Friday’s ruling. As reported by CNBC, a company spokesperson stated, “We respectfully disagree with the court’s decision. Another federal court has already held the government’s parallel designation unlawful.”
It remains uncertain whether Anthropic will pursue a rehearing or elevate the dispute to the Supreme Court, leaving this emerging AI battleground divided across two conflicting federal courts.