A federal appeals court has upheld the Pentagon's decision to label AI firm Anthropic as a supply chain risk. This 2-1 ruling allows the Department of Defense to purge Claude models from its systems and prohibit contractors from using them.
The DC Circuit's 2-1 ruling on Claude's removal
The U .S. Court of Appeals for the District of Columbia Circuit ruled 2-1 in favor of the Pentagon, confirming the government's authority to designate Anthropic as a supply chain risk.. According to reporting by the Associated Press, this decision empowers the Department of Defense to bar the use of Anthropic's products for official work and ensures that contractors cannot utilze these tools when fulfilling defense contracts.
Sean Parnell, the top spokesman for the Pentagon, stated in a social media post that the ruling "completely validates" the department's position. while the ruling is a significant blow to the AI lab, it is important to note that Anthropic's products are not barred from the federal government as a whole,only from the specific workflows of the Department of Defense.
How Anthropic's safety guardrails became a national security risk
The core of the legal dispute centers on the very safety features Anthropic built into its Claude models. Judge Gregory G. Katsas wrote in the majority opinion that the Pentagon had ample support for its actions because Anthropic admits to encoding restrictions into Claude to prevent the model from performing certain tasks.
The Department of Defense argued that these private restrictions could be manipulated to block the AI from performing national-security functions that the government deems contractually authorized and necessary. As the report says, the court found that the definition of a supply chain risk depends on what the company does—in this case, restricting its own software—rather than the intentions behind those restrictions.
The clash between Dario Amodei and the Trump-Hegseth administration
This legal battle is the culmination of a dispute that began in February, when President Donald Trump and Defense Secretary Pete Hegseth accused Anthropic of endangering national security. Anthropic CEO Dario Amodei resisted the government's pressure, citing concerns that the company's AI products could be weaponized for autonomous armed drones or used for mass surveillance.
This conflict reflects a growing tension between the "AI safety" movement in Silicon Valley and the operational requirements of the U.S. military. The majority opinion was authored by Judges Katsas and Neomi Rao, both of whom were nominated by Donald Trump, while the sole dissenter, Judge Karen LeCraft Henderson, was a nominee of George H. W. Bush.
The contradiction between the DC ruling and the San Francisco court's findings
The DC Circuit's decision stands in contrast to a separate case filed in San Francisco. In that instance, a federal judge ruled that the Pentagon had acted illegally by punishing the AI company for criticizing the government's views on AI usage. However, the DC judges clarified that a "bad motive" is not required to support a supply chain risk designation, effectively decoupling the company's political speech from its technical restrictions.
Several critical questions remain unresolved following this ruling. While an Anthropic spokesperson stated the company is "considering all options, including further revew," it remains unclear if the lab will appeal to the Supreme Court. Additionally, the source does not specify which specific "national-security functions" the Pentagon believes Claude was preventing, nor does it detail how the Department of Defense intends to replace the Claude workflows it has already begun removing.
Comments 0