ABC News: Pentagon gives Anthropic Friday deadline over military AI use

The Pentagon has reportedly given Anthropic until Friday to accept its terms for military use of the company’s AI, or face possible government action. The dispute puts a sharp question on the table: how much control should an AI supplier retain over how its models are used by the military?

Anthropic Watch analysis

What happened

ABC News reports, citing a Pentagon official and a source familiar with the matter, that the department is demanding Anthropic agree to its terms by Friday. The Pentagon official told ABC that Defence Secretary Pete Hegseth threatened to invoke the Defense Production Act to compel compliance, and to designate Anthropic a “supply chain risk”. Those are reported threats, not actions ABC says have already been taken.

A source familiar with a Tuesday meeting between Hegseth and Anthropic chief executive Dario Amodei told ABC that Amodei set out the company’s limits, including autonomous weapons and mass domestic surveillance. The Pentagon official said the dispute was not about those uses and that the department follows the law. Anthropic said the two sides were continuing good-faith talks about its usage policy and how the company could support national security within the limits of what its models can reliably and responsibly do.

Why it matters

The Pentagon has been using Anthropic technology in a reported $200 million pilot programme. A threat to compel a supplier to accept government terms, or to cut it off from Pentagon business and partners, would make this more than a disagreement over contract language. It would test where a company’s usage limits end and government authority over a technology supplier begins.

The competing accounts matter. ABC’s sources describe an ultimatum and threats; the Pentagon disputes the connection to the red lines Anthropic reportedly raised. The deadline is approaching, but the supplied report does not establish that either threatened measure has been carried out.

Our read

This is a consequential standoff, not a settled change in Pentagon policy. The practical distinction to keep in view is between a threat, an actual designation and a legal order: they would have different consequences, and only the first is reported here. The argument over who sets limits on military AI deserves attention precisely because the tools are already part of a government pilot.

What to watch

• Whether Anthropic and the Pentagon reach an agreement before the reported Friday deadline.
• Whether the Pentagon takes either of the actions ABC says was threatened.
• What the parties say about the terms of military use and Anthropic’s usage limits.

Discussion spark: If a government says it needs broad access to AI for national security, should the supplier be able to refuse particular uses, or should that decision belong to elected officials and the military?

Anthropic Watch is independently operated by WittyWires. It is not affiliated with, endorsed by, or operated by Anthropic.