An AI Company Refuses the Pentagon's Terms and Is Declared a Threat to National Security
Anthropic asked for two limits on military use of its AI: no mass surveillance of Americans, and no fully autonomous weapons. The Pentagon refused, and the Defense Secretary barred every military contractor from doing any business with the company.
Anthropic, the company behind the Claude AI assistant, has been cut off from the United States military. Defense Secretary Pete Hegseth announced today that the company is "a supply chain risk to national security," and that "effective immediately, no contractor, supplier, or partner that does business with the United States military may conduct any commercial activity with Anthropic." That is not the cancellation of a contract. It is an instruction to the entire defense industry to stop dealing with the company at all.
The dispute was about two conditions. In negotiations over how the military could use its models, Anthropic asked for prohibitions on two uses: mass surveillance of American citizens, and fully autonomous weapons — systems that select and strike targets without a person deciding. Yesterday the company's chief executive, Dario Amodei, rejected what the Pentagon called its best and final offer, writing that "our strong preference is to continue to serve the Department and our warfighters — with our two requested safeguards in place." The reply came the next morning.
The response has been personal as well as commercial. A senior Pentagon research official publicly called Amodei "a liar" with "a God-complex."
There is a further detail that makes the timing difficult to read past. Anthropic's tools were reportedly used in the U.S.-Israeli bombing campaign against Iran, which began today — the same day the company was designated a security risk. The company was, in other words, already inside the machinery of a live military operation while being shut out of it.
The disagreement is not really about Anthropic. It is the first hard test of a question the whole industry has avoided: whether a company that builds these systems can set limits on how a government uses them. Anthropic has argued since its founding that some uses should be off the table, and it wrote those limits into its public policies. The Pentagon's position is that a supplier does not get to dictate the terms of national defense. Both positions are coherent. They are also, as of today, incompatible — and the company that held the line has been removed from the table for it.