Pentagon officials sent Anthropic best and final offer for military use of its AI amid dispute, sources say

Pentagon officials on Wednesday night sent Anthropic their best and final offer in negotiations for use of the company’s artificial intelligence technology, just ahead of a government-imposed deadline, according to sources familiar with the discussions. 

It was unclear whether the offer substantially changed what the government has been seeking from the AI startup, or whether the company had agreed. 

Defense Secretary Pete Hegseth set a deadline of Friday evening for the company to grant all lawful use for its AI technology or face the loss of its business with the U.S. military, sources familiar with the situation told CBS News. 

Spokespeople for the company didn’t immediately respond to a request for comment Thursday morning. 

A senior Pentagon official said Thursday Anthropic will face not just the loss of business but being labeled a supply chain risk. 

Pentagon officials are also considering invoking the Defense Production Act to make Anthropic adhere to what the military is seeking, which is full control of its AI technology for use in military operations, sources told CBS News. 

The company was awarded a $200 million contract by the Pentagon in July to develop AI capabilities that would advance U.S. national security.

Anthropic has repeatedly asked defense officials to agree to guardrails that would restrict its AI model, called Claude, from conducting mass surveillance of Americans, sources said. 

Trump officials noted that this sort of surveillance is illegal and the Pentagon follows the law. The officials also said the military is simply asking for a license to use the AI strictly for lawful activities.

Anthropic’s CEO, Dario Amodei, also wants to ensure Claude is not used by the Pentagon for final targeting decisions in military operations without any human involvement, one source familiar with the negotiations said. Claude is not immune from hallucinations and not reliable enough to avoid potentially lethal mistakes, like unintended escalation or mission failure without human judgment, the person said. 

In a meeting at the Pentagon on Tuesday morning, Hegseth gave Amodei until the end of this week to give the military a signed document that would grant full access to its artificial intelligence model, according to sources familiar with the matter.