Trump banned Anthropic, OpenAI cut a deal, and nobody is completely wrong here

You probably use AI every day now whether you realize it or not and this week, the fight over who controls it just got a whole lot more interesting.

The Trump administration ordered every federal agency to immediately stop using products made by Anthropic, the company behind the Claude AI.

Defense Secretary Pete Hegseth went further and designated Anthropic a national security supply chain risk, meaning no military contractor or partner can do business with them either.

The ban came with a six-month phaseout window. Within hours of the announcement, rival AI company OpenAI announced it had cut a deal with the Pentagon to fill the gap.

This is a genuinely fascinating story and the truth is that both sides have a point.

What Anthropic Was Saying

Anthropic had a contract with the Pentagon worth up to $200 million. The dispute came down to two specific things Anthropic refused to allow its AI to be used for: mass domestic surveillance of American citizens and fully autonomous weapons systems, meaning weapons that fire without a human making the final call.

Anthropic CEO Dario Amodei said the company tried in good faith for months to reach an agreement and made clear it would support all lawful uses of AI for national security except those two things.

His argument was straightforward: today’s AI is simply not reliable enough to be trusted with autonomous weapons, and using it for mass surveillance of Americans is a violation of fundamental rights.

Those are not fringe positions. Most AI researchers and ethicists would agree with both of them. OpenAI CEO Sam Altman said on Friday morning that he shares those exact same red lines. His words. He said the prohibitions on domestic mass surveillance and human responsibility for the use of force are two of OpenAI’s most important safety principles.

He just also cut a deal with the Pentagon anyway, saying the Department of War agreed to reflect those principles in law and policy. So OpenAI got the government to agree in writing. Anthropic could not get there. That is the part worth paying attention to.

What the U.S. Government Was Saying

The Pentagon’s argument is also reasonable on its face.

Defense Department officials said federal law and Pentagon policy already prohibit AI from being used for domestic mass surveillance or fully autonomous weapons.

Their position was that they do not need a private company putting additional restrictions on top of existing law, because that would set a precedent where every AI contractor gets to negotiate their own carveouts for how the military operates.

Emil Michael, the Pentagon’s undersecretary for research and engineering, accused Amodei of having a “God-complex” and wanting to personally control the U.S. military.

Trump’s Truth Social post was considerably less diplomatic, calling Anthropic “leftwing nut jobs” trying to force the government to obey their terms of service instead of the Constitution.

Trump: “THE UNITED STATES OF AMERICA WILL NEVER ALLOW A RADICAL LEFT, WOKE COMPANY TO DICTATE HOW OUR GREAT MILITARY FIGHTS AND WINS WARS! That decision belongs to YOUR COMMANDER-IN-CHIEF, and the tremendous leaders I appoint to run our Military. 
 
The Leftwing nut jobs at… pic.twitter.com/DOxVxCJWFW

— Aaron Rupar (@atrupar) February 27, 2026

The underlying concern is legitimate even if the framing is aggressive. A private company telling the military how it can and cannot use a tool in a combat or intelligence context is genuinely unusual.

Pentagon contractors do not typically get to dictate use cases. As one defense industry expert put it this week, if you negotiated use restrictions for every contract you would never get anything done.

The OpenAI Angle

The most interesting part of this whole story is what Sam Altman did on Friday. He spent the morning on CNBC saying he shares Anthropic’s red lines. Then he announced OpenAI had struck a deal with the Pentagon that includes those exact protections written into the agreement. He essentially said: I agree with Anthropic on the principles, I just found a way to make it work.

Tonight, we reached an agreement with the Department of War to deploy our models in their classified network.

In all of our interactions, the DoW displayed a deep respect for safety and a desire to partner to achieve the best possible outcome.

AI safety and wide distribution of…

— Sam Altman (@sama) February 28, 2026

Whether that deal holds up in practice is a different question. But it does suggest that Anthropic’s position was not the obstacle. The negotiations were.

Why This Matters Beyond AI

Anthropic is valued at $380 billion and is planning to go public this year. The Pentagon contract was only worth up to $200 million, which is a relatively small slice of $14 billion in revenue.

But being designated a national security supply chain risk is a different kind of problem. It means military contractors who currently use Anthropic products for their own work now have to cut ties with the company entirely, which is a much bigger hit than losing one government contract.

Anthropic said Friday it would challenge the designation in court, arguing the government overstepped its authority and that the designation can only apply to the Pentagon contract, not to how contractors use Claude for their other customers.

This fight is not over and it raises questions that go well beyond Anthropic or OpenAI.

Who decides how AI gets used in war? Who has the authority to put limits on it? What happens when the most powerful technology in the world is being deployed by the most powerful military in the world and the two sides cannot agree on the rules?

Nobody has a clean answer to any of those questions yet but the fact that we are asking them right now, out loud, in a very public and very contentious way, is probably the most important thing about this story.

Join The Chase