Hegseth considering severe penalty on Anthropic as negotiations stall

By Steve Goldstein

U.S. Defense Secretary Pete Hegseth is reportedly considering a severe penalty against Anthropic, the maker of Claude.

Defense Secretary Pete Hegseth is close to putting artificial-intelligence company Anthropic on a "supply-chain risk" list that would be a devastating blow to its finances.

The Pentagon in a statement said the U.S. government was "reviewing" its relationship with Anthropic, the maker of the Claude large language model.

"The Department of War's relationship with Anthropic is being reviewed. Our nation requires that our partners be willing to help our warfighters win in any fight. Ultimately, this is about our troops and the safety of the American people," said Chief Pentagon Spokesman Sean Parnell in a statement.

A senior official said that the Pentagon was starting to view Anthropic as a supply-chain risk and may require all vendors and contractors certify that they don't use any of the company's models.

Axios was first to report the development.

Merely losing the Pentagon contract would be less of a financial blow, given it's valued at up to $200 million. Anthropic says it now generates revenue at an annual rate of $14 billion.

The Pentagon is a user of Anthropic's Claude and reportedly used it, through a Palantir (PLTR) product, in the raid that resulted in the apprehension of Venezuelan President Nicolás Maduro. Axios reported Claude is the only AI model available in the Defense Department's classified systems.

The issue is with Anthropic's negotiations over how its products can be used, drawing a line at it being used to spy on Americans en masse or to develop weapons that fire with no human involvement, according to Axios. The report said the government is concerned about "gray areas" as well as the prospect of having to negotiate individual use cases.

An Anthropic spokesman was quoted as saying that there are "hard limits around fully autonomous weapons and mass domestic surveillance" but denied that it had objected to the Venezuela operation. Anthropic did not respond to a message from MarketWatch.

Rival AI models from OpenAI, Alphabet's (GOOGL) Google and SpaceX's xAI are used by the Pentagon, but only in unclassified settings. "Other AI companies are working collaboratively with the Pentagon in good faith to ensure their models can be used for all lawful purposes," the senior official said. The official added the government was confident those models would soon be available in classified settings.

Anthropic last week raised funds at a $380 billion valuation and, along with OpenAI and SpaceX, is one of the most hotly awaited initial public offerings from companies that are still private.

-Steve Goldstein

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.

(END) Dow Jones Newswires

02-17-26 1052ET