Pentagon and AI company Anthropic face off over military use of AI

A major dispute is heating up between the U.S. military (the Pentagon) and Anthropic, one of the leading companies building powerful artificial intelligence tools like the Claude model. The fight is about how much freedom the military should have to use Anthropic’s AI in warfare and other operations.

A critical deadline looms: both sides must reach an agreement by 5:01 p.m. Eastern Time (that’s 2201 GMT) on Friday, or face serious fallout.

Here’s what’s happening in simple terms:

The Pentagon signed a deal worth up to $200 million with Anthropic (and similar agreements with other AI firms like OpenAI and Google) to help the military use advanced AI for things like analysing data, planning operations, and improving decision-making. Anthropic was the first to provide its technology for handling classified (secret) information through Amazon’s secure cloud system.

But trouble started when the military pushed to remove extra safety restrictions that Anthropic built into its AI. These “guardrails” block the technology from being used for things like fully autonomous weapons (AI systems that make kill decisions without a human in the loop) or mass surveillance of American citizens at home. Anthropic says these limits are essential because today’s AI isn’t reliable enough for such high-stakes or ethically sensitive uses, and allowing them could harm democratic values.

The Pentagon insists that as long as something is legal under U.S. law, the military should be allowed to use the AI without extra company-imposed rules. Officials argue they have no interest in illegal mass spying on Americans or creating weapons that act completely on their own without human oversight. Pentagon spokesperson Sean Parnell posted on X (formerly Twitter) that the department doesn’t want autonomous weapons without human involvement or domestic surveillance, which is already against the law.

The disagreement escalated this week: Secretary of War Pete Hegseth met with Anthropic CEO Dario Amodei, and the Pentagon sent over new contract wording overnight. Anthropic rejected it, saying it still lets safeguards be ignored whenever desired. Amodei posted a statement saying his company “cannot in good conscience” agree to the demands, even under pressure.

A statement from Anthropic CEO, Dario Amodei, on our discussions with the Department of War.https://t.co/rM77LJejuk
— Anthropic (@AnthropicAI) February 26, 2026

If no deal is reached by Friday, the Pentagon has threatened to:

Cancel the contract.
Label Anthropic a “supply chain risk” (a label usually used for companies from hostile countries like China or Russia), which could force defense contractors (like Lockheed Martin) to stop using Anthropic’s AI.
Possibly use the Defense Production Act, a law that lets the government force companies to prioritize military needs in emergencies.

This isn’t just about one company – it’s a bigger test of how AI will be used in future wars. Former acting Defense Secretary Chris Miller called it a “shot across the bow” about AI on the battlefield. Some lawmakers, like Democratic Senator Elissa Slotkin, support keeping human oversight on weapons and oppose any mass surveillance of citizens.

Anthropic, founded by people who left OpenAI over concerns about rushing AI development too fast, has built its reputation on responsible AI. The outcome could affect whether other AI companies feel comfortable working with the military, and how quickly powerful AI gets integrated into defense without strong ethical checks.

With the deadline just hours away, the world is watching to see if the two sides can compromise or if this feud leads to a major break in U.S. military-AI partnerships.

BREAKING: 300+ Google & OpenAI employees just did something WILD.
In 8 hours, the Pentagon's ultimatum expires.
Instead of their companies giving in, employees just made it impossible to comply.
Either Anthropic removes all AI safeguards for military use, or face the Defense… https://t.co/CY6c5HHrv3 pic.twitter.com/7VBhE4hNwC
— Shruti (@heyshrutimishra) February 27, 2026

This is a huge opportunity for Europe. Welcome Anthropic with open arms. Roll out the red carpet. Visa for all employees.
Europe already controls the AI hardware bottleneck through ASML. Add the world's leading AI safety lab and you have the foundations of an AI superpower. https://t.co/E9FWtI8yDT
— Rutger Bregman (@rcbregman) February 27, 2026

With Reuters, edited by g.g.

Caption: Screenshot from video showing Anthropic CEO Dario Amodei at the New York Times DealBook Summit on 4 December 2025 [@AnthropicAI]