美国上诉法院裁定五角大楼可继续禁止Anthropic参与军事合同

A federal appeals court has handed the Trump administration a victory in its fight with Anthropic.

In a new appeals court judgement A U.S. appeals court has allowed the Pentagon to continue blocking Anthropic from military contracts for now over its “national security risk” designation. handing a victory to the Trump Administration.

• The D.C. Circuit said Claude's built-in restrictions can qualify as a supply-chain risk because they blocked government-requested tasks and created uncertainty during an overseas military operation.
• The majority did not rely on Anthropic retaining remote control after deployment, because the Pentagon had clarified that its designation did not rest on that premise.
• Instead, its reasoning turns on whether Claude's encoded safety rules can make the model refuse military tasks when the Pentagon considers reliability operationally critical.
• The majority also accepted Anthropic's AI-safety advocacy as protected speech, but said the exclusion flowed from rejected contract terms rather than retaliation for that advocacy.
• But this is not a nationwide judgment that every federal agency can blacklist Anthropic. Judge Rita Lin's separate California ruling struck down the broader §3252 designation and actions including the government-wide prohibition and the attempt to prevent defense contractors from doing unrelated business with Anthropic.

So, oddly, both judgments can exist at the same time because they concern different statutory powers. The D.C. Circuit itself stressed that §4713 has a substantially broader definition of "supply chain risk" than §3252 and expressly said it had "no quarrel" with the California court's conclusion that Anthropic lacked the bad motive required under the narrower law.