Anthropic vs US Government: DC court showdown over AI company blacklisting

A federal appeals court in Washington, D.C., is scheduled to review arguments on Tuesday regarding Anthropic's lawsuit concerning its blacklisting by the Department of Defense.

This marks the newest development in the ongoing dispute between the Pentagon and one of the nation's prominent AI enterprises.

The US Department of Justice, representing the DOD, and Anthropic are allotted 15 minutes each to argue their case before a trio of circuit judges, as per an order issued earlier this month. Judge Karen Henderson, Judge Gregory Katsas, and Judge Neomi Rao will deliberate before providing a written decision.

The session will commence at 9:30am ET on Tuesday.

Anthropic initiated legal action against Defense Secretary Pete Hegseth and the DOD in March after the agency labelled the AI startup as a supply chain risk, implying it could jeopardise US national security.

Historically such labels target foreign threats, obligating defense contractors to certify they won't utilise Anthropic's Claude models in military collaborations.

The classification came post enduring intense dialogues between Anthropic and the DOD, which disintegrated.

The DOD sought unrestricted access to Anthropic's models for all legal operations, whereas Anthropic wanted assurances against its technology being utilised for autonomous weapons or domestic mass surveillance.

As negotiations faltered, Hegseth excluded Anthropic and criticised the company on social platforms. Anthropic CEO Dario Amodei remarked the company saw "no alternative" but to contest the supply chain risk designation in court.

The DOD still employs Anthropic's models in military operations concerning Iran, and President Donald Trump suggested to CNBC last month that a settlement between the DOD and Anthropic is "feasible."

The appeals court refused Anthropic's plea to temporarily halt the designation in April, maintaining its effect while the legal proceedings unfold. However, due to potential "irreparable harm" Anthropic might face during the trial, judges agreed to accelerate the case as mentioned in an order.

In a brief prior to Tuesday's arguments, the government contended that Anthropic could "integrate limitations" into its model, posing a "critical national-security threat." Hegseth stated that Anthropic "compromised the significant trust needed for a sustainable relationship," especially as the company might "adapt its model to impose its ethical and policy considerations on the military's technological use," as per the brief.

Anthropic, in a different brief, countered the claim that it could encode restrictions in future models was unfounded and offered "no justification" for a supply chain risk label. They further argued that Hegseth and the DOD breached both the Constitution and established procedures.

"The Court ought to deem the designation unlawful," Anthropic's legal representatives asserted.

In addition to its Washington, D.C. lawsuit, Anthropic instigated a separate but connected lawsuit in federal court in San Francisco. The DOD utilised two distinct designations to mandate its supply chain risk action, necessitating dual court trials.

Anthropic received a temporary restraining order in its San Francisco proceedings, enabling government entities aside from the DOD to avail Anthropic's models while the court process persists.

"The existing statute bears no support for the dystopian idea that an American corporation may be labelled a potential enemy and saboteur of the US for opposing government actions," the judge ruled.