Pentagon Rejects Anthropic Appeal, Paving Way for Court Ruling on AI Security Powers

Defense Secretary Pete Hegseth signed off on a decision in early June that kept the AI company Anthropic labeled a national security risk, a move that now forces a federal appeals court to weigh whether the Pentagon can designate an American technology firm as a supply chain threat for the first time.

The Pentagon notified the D.C. Circuit Court of Appeals on Thursday that Hegseth had denied Anthropic’s request for an internal reconsideration, closing the administrative path and clearing the way for a three-judge panel to hear the case.

The dispute began over disagreements about how Anthropic restricts the use of its AI models, particularly in areas like domestic surveillance and autonomous weapons. The Pentagon had designated the company a supply chain risk, a label that historically applied only to foreign companies with ties to U.S. adversaries. Anthropic sued, challenging the scope of the Defense Department’s authority over domestic companies.

During oral arguments in March, judges on the panel, including two appointees of former President Donald Trump, indicated they were likely to issue a ruling that would grant the executive branch broad power to label domestic companies as supply chain risks. However, the judges also suggested that subsequent directives barring military or other government agencies from using a company’s products could still face legal challenges.

In his June 3 decision, Hegseth wrote that “pre-deployment risks with the Covered Entity’s products and services, the loss of trust, and other risk factors … were and remain sufficient to support the prior Determination.” The Pentagon urged the court to decide the urgent questions raised in the case now that the internal review process had concluded.

Hegseth also clarified that the initial designation rested on a loss of trust and other pre-deployment risks associated with Anthropic’s Claude AI model, rather than on the company’s supposed ability to unilaterally manipulate the model in real time after deployment. Anthropic has argued that it does not actually possess that capability and that Hegseth initially relied on a misunderstanding.

Legal Questions Now in Judges’ Hands

Judges had raised concerns last month that the courts might have to wait for the Pentagon to handle Anthropic’s reconsideration request before they could take up the case. With that request now denied, the panel is expected to decide the novel legal questions surrounding the Pentagon’s designation powers over domestic technology companies. The ruling could set a precedent for how the federal government labels and restricts AI firms it deems security risks.