Although ChatGPT is dominating the consumer space with hundreds of millions of users, Anthropic"s models are quite popular in specialized environments, too. However, the company is feuding with the U.S. Department of Defense over matters of national security and the use of its technology in related efforts.
An exclusive report from Axios claims that Defense Secretary Pete Hegseth is very close to cutting ties with Anthropic and branding it as a "supply chain risk", which basically means that any company that wants to work with the U.S. military will first have to ensure that it is not working with Anthropic in any capacity either.
The fallout stems from reports that Anthropic"s flagship Claude model was utilized in the recent arrest of incumbent Venezuelan President Nicolás Maduro by the United States. It is also notable that Claude is the only AI model that has the operational green light in classified U.S. military systems.
However, Anthropic is not too happy with its AI models being used for any and all purposes. The firm has been negotiating with the Pentagon that it does not want its tools to be used for the mass surveillance of Americans or to be capable of firing weapons without human intervention.
The Pentagon is not happy with the situation, noting that there are a lot of gray areas where Anthropic"s demands are unfairly restrictive. As the department considers cutting off Anthropic altogether, it has promised hell to pay for the firm, with Hegseth quoted as saying, "It will be an enormous pain in the ass to disentangle, and we are going to make sure they pay a price for forcing our hand like this."
On the other side of the fence, Anthropic is still hoping for a favorable turn in the negotiations, which it is carrying out in "good faith". However, that outcome seems unlikely as the Pentagon is reportedly also in early stages of discussions with competitors like OpenAI, Google, and xAI, who may be more willing to bend to the "all lawful use" standard set for AI models by the Pentagon.