Anthropic Could Face 'Supply Chain Risk' Tag As Pentagon Weighs Cutting Ties: Report

Defense Secretary Pete Hegseth is reportedly contemplating ending the Department of War’s association with AI firm Anthropic.

AI Safeguards Clash With Pentagon

The Department of War’s relationship with Anthropic is under scrutiny, Chief Pentagon spokesman Sean Parnell told Axios on Monday.

Another high-ranking Pentagon official told the publication that the company could soon be labeled as a “supply chain risk.” He also stated that it would be "an enormous pain" to disentangle from the arrangement and added that they would ensure the company "pays a price" for forcing their hand.

If Anthropic were labeled a supply chain risk, companies working with the Pentagon would have to certify they don't use its Claude AI, which could be challenging.

Anthropic’s AI model, Claude, is the only one integrated into the military’s classified systems and is extensively utilized in various business applications. Axios had earlier reported that the AI model was also used in the capture of the former Venezuelan President, Nicolás Maduro, in January.

Despite the ongoing discussions, Anthropic is prepared to relax its terms of use but remains adamant about preventing its tools from being used for mass surveillance of Americans or for developing autonomous weapons. The Pentagon, however, deems these conditions excessively restrictive.

On previous occasions, the Trump administration has been accused of allegedly using Palantir Technologies Inc. (NASDAQ:PLTR)  to spy on U.S. citizens, a claim that the company repeatedly denied.

The Department of War and Anthropic did not immediately respond to Benzinga‘s requests for comment.

Amodei Warns On Chips, AI Spending

This development follows the company’s CEO, Dario Amodei, previously criticizing NVIDIA Corp (NASDAQ:NVDA) and the Trump administration for their dealings on chip sales to China. Amodei warned that the issue goes beyond the AI race, arguing the chips pose a national security risk and likening their sale to "selling nuclear weapons to North Korea."

This report of the Anthropic-Pentagon rift also comes on the heels of the company pledging $20 million to Public First Action (PFA), a bipartisan group working to advance AI governance and public education, promote transparent safeguards, strengthen federal oversight, and enforce export controls to preserve U.S. leadership in AI.

Meanwhile, the company’s CEO expressed caution about going overboard with the AI investment. Amodei said it's unclear how quickly trillions in revenue would materialize, estimating it could take one to five years, and warned that investing massive capital too early could be ruinous if projections miss the mark.

Notably, Anthropic’s Pentagon contract is valued at up to $200 million out of its $14 billion in annual revenue, as per the publication.

Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors.

Image via Shutterstock