A recent hearing in the U.S. Court of Appeals for the District of Columbia Circuit revealed a split among judges regarding the Pentagon's legal stance on Anthropic, a technology company accused of posing a national security threat. This case has significant implications for the governance of AI technologies, particularly in military applications.
The dispute centers on comments made by Defense Secretary Pete Hegseth, who labeled Anthropic as a supply chain risk after the company expressed ethical concerns about AI use in warfare. During the proceedings, Judge Karen LeCraft Henderson sharply criticized the Pentagon's portrayal of Anthropic, stating, "To me, this is just a spectacular overreach by the (Defense) Department." Her remarks suggest she may lean towards supporting Anthropic in the appeal.
In contrast, Judge Neomi Rao, nominated by former President Donald Trump, examined the limits of the court's authority to question Hegseth's judgment. Rao indicated that the Pentagon's concerns about the reliability of AI models in military operations were based on a broader context of risk management. She noted, "I take the secretary to be making more general points than the ones that you’ve identified."
Anthropic has initiated lawsuits in both Washington D.C. and San Francisco following the Pentagon's designation. The company argues that the Defense Department's actions have caused irreparable harm by stigmatizing it and creating obstacles for its technology to be used by federal agencies. Importantly, Anthropic has clarified that its legal actions aim not to force government contracts but to challenge the validity of the risk designation itself.
In a related ruling, a federal judge in San Francisco recently sided with Anthropic, blocking the Pentagon's classification of the company as a supply chain risk. This ruling stands in stark contrast to the D.C. Circuit's earlier decision to deny Anthropic's request for an injunction that would have paused the Pentagon's actions while the appeal was ongoing.
The Pentagon's argument hinges on the claim that Anthropic's AI model, Claude, could present significant risks if deployed within classified military networks. Justice Department attorney Sharon Swingle emphasized the potential dangers, stating, "It’s undisputed that the failure of the model in active military operations could have catastrophic national-security consequences and put service members’ lives at risk."
https://www.youtube.com/watch?v=3Tj4NxpBgFI
Anthropic's attorney, Kelly Dunbar, countered these assertions by claiming that Hegseth's designation strayed from established procedures and constitutional limits. Dunbar argued that this represents an unprecedented use of national security measures against an American company, suggesting that the decision was driven by contract negotiations rather than legitimate security concerns.
As the case progresses, the implications for AI technology in military contexts remain substantial. If the court rules in favor of Anthropic, it could set a precedent that restricts the Pentagon's ability to impose such designations on tech companies unilaterally, potentially encouraging a more cooperative environment for AI innovation in defense. On the other hand, a ruling favoring the Pentagon may strengthen stringent oversight on AI technologies, particularly in sensitive applications.
The judges are expected to deliberate further before issuing a ruling, though no timeline for their decision has been provided. With high stakes involved, the outcome of this legal battle will likely have repercussions beyond the courtroom, affecting future interactions between technology firms and government agencies at the intersection of national security and emerging technologies.
