Pentagon gives Anthropic deadline over military AI terms
The Defense Department gave Anthropic until Friday to accept its terms for military AI use or face forced compliance, ABC News reported. The dispute centers on Anthropic’s requested limits on mass domestic surveillance and fully autonomous weapons.

Key takeaways · 4
- 01
The Pentagon set a Friday deadline for Anthropic to accept its terms, with forced compliance threatened if it did not.
- 02
The dispute concerns Anthropic’s requested limits on mass domestic surveillance and fully autonomous weapons.
- 03
The Pentagon had been using Anthropic technology in a $200 million pilot program.
- 04
Anthropic says a supply-chain-risk designation would affect Claude used directly on Defense Department contract work, not all customer use.
A deadline and potential penalties
ABC News cited a Pentagon official and a source familiar with the matter in reporting the ultimatum.[1] A source familiar with a meeting said it was respectful but ended with a warning that the Pentagon would punish Anthropic if it rejected the terms.[1] Anthropic said Hegseth had shared that he was directing the Department of War to designate the company a supply-chain risk.[3]
The disagreement over use limits
Anthropic says it supports lawful national-security uses of AI, with two narrow exceptions, and that those two use cases had never been included in its Department of War contracts.[2][3] The company says frontier AI systems are not reliable enough to power fully autonomous weapons.[2] The Pentagon official, meanwhile, said the dispute was not about the department using AI for mass surveillance or autonomous weapons and said the government, as the end user, was responsible for ensuring legal use.[1] The accounts frame the conflict around who sets limits on permitted uses and who is accountable for them.
Claude already supports national-security work
Anthropic says it proactively deployed models to the Department of War and the intelligence community.[2] The company says Claude is used across the Defense Department and other national-security agencies for intelligence analysis, modeling and simulation, operational planning, and cyber operations.[2] The Pentagon had also been using Anthropic technology in a $200 million pilot program.[1] Anthropic says it held good-faith talks with the Pentagon about its usage policy and continued support for national-security work.[1] It says its restrictions had not affected a government mission, to its knowledge.[3]
What a designation would cover
Anthropic says a supply-chain-risk designation would affect a defense contractor’s use of Claude as a direct part of Defense Department contract work, rather than every use by customers that hold such contracts.[3][4] It says access through the Claude API, claude.ai, and its products would be unaffected.[3] Anthropic said it received a letter on March 4 confirming the designation and would challenge any such designation in court.[3][4] It also said it would provide models and engineering support to the Defense Department and national-security community at nominal cost during a transition, for as long as necessary and permitted.[4]
For professionals, the key issue is how an AI-use policy and any supply-chain designation apply to particular workflows, not simply whether a vendor is available. Teams supporting defense work should clarify which tools are used directly on contract work and plan for continuity if permitted use changes.
Why it matters
Put this to work — one session a day, built for your industry.
Create a free account for a daily session — eight questions and one real-work challenge, on the news that affects your role.
Start freeHow this developed
19 October 2026
Pentagon gives Anthropic deadline over military AI terms
Sources
- Pentagon gives Anthropic ultimatum on AI technology: Sources - ABC Newsabcnews.com
- Dario Amodei on the Department of War discussions \\ Anthropicanthropic.com
- Statement on the comments from Secretary of War Pete Hegseth \\ Anthropicanthropic.com
- Where things stand with the Department of War \\ Anthropicanthropic.com