Trump Bans Anthropic from Federal Contracts

The AI company refused to let the Pentagon use Claude for mass domestic surveillance or autonomous weapons. Now it's blacklisted.

TL;DR:

  • Trump ordered all federal agencies to stop using Anthropic (the company behind Claude) effective immediately, with a six-month phaseout for agencies already using it
  • Defense Secretary Hegseth declared Anthropic a "supply chain risk", the same designation usually reserved for foreign adversaries like Huawei
  • The $200 million Pentagon contract is dead. Anthropic was the first AI company cleared for classified military use, now it's banned
  • Why? Anthropic refused to remove guardrails preventing Claude from being used for mass domestic surveillance or fully autonomous weapons
  • OpenAI is already in talks with the Pentagon, but Sam Altman says he shares Anthropic's "red lines"

The Showdown

It came down to a Friday deadline. At 5:01 p.m. on February 27, Defense Secretary Pete Hegseth made the break official.

"Anthropic's stance is fundamentally incompatible with American principles," Hegseth posted on X. He directed the Pentagon to designate Anthropic a "supply-chain risk to national security": meaning no contractor, supplier, or partner that does business with the U.S. military can conduct any commercial activity with Anthropic.

The announcement came less than 24 hours after Anthropic CEO Dario Amodei said he "cannot in good conscience accede" to the Pentagon's demands.

That's corporate-speak for: we won't help you spy on Americans or build killer robots.

What Anthropic Refused to Do

The Pentagon wanted two things Anthropic wouldn't give:

  1. Mass domestic surveillance: Using Claude to analyze data collected on American citizens at scale
  2. Fully autonomous weapons: AI systems that identify and kill targets without human approval

Hegseth demanded Anthropic allow the military to use Claude "for all lawful purposes." Anthropic's position: just because something is legal doesn't mean we'll enable it.

The Pentagon offered a compromise: missile defense only. Anthropic countered with "defensive uses only." Neither side budged on the core issues.

According to Anthropic, the contract language they received from the Pentagon "made virtually no progress on preventing Claude's use for mass surveillance of Americans or in fully autonomous weapons."

What "Supply Chain Risk" Actually Means

This designation isn't symbolic. It's the same label slapped on Huawei and ZTE, companies the U.S. considers security threats due to foreign government ties.

For Anthropic, it means:

  • All federal agencies must immediately stop using Anthropic products
  • Agencies already deployed have six months to phase out
  • Every Pentagon contractor must certify they don't use Claude in their workflows
  • Companies doing business with Anthropic can't do business with the military

That's a business-killing ultimatum. Defense contractors are going to have a choice: the Pentagon or Anthropic. For most, it's not even close.

The $200 Million Contract Is Gone

Last summer, the Pentagon awarded Anthropic, Google, OpenAI, and xAI contracts each worth up to $200 million. Anthropic was the first to be cleared for classified use after defense officials considered it the most advanced and secure model for sensitive military applications.

That's done now. The contract is terminated.

One Pentagon official reportedly called Amodei's position a "God-complex": the implication being that refusing government demands is arrogance, not principle.

OpenAI Is Waiting in the Wings

Sam Altman is threading a needle. Hours after Anthropic's ban, OpenAI confirmed it's in talks with the Pentagon about moving ChatGPT into classified systems.

But Altman told staff he shares Anthropic's "red lines": no mass surveillance, no autonomous weapons. OpenAI employees reportedly voiced support for Anthropic's position during an all-hands meeting.

Here's the problem: if OpenAI holds the same lines, the Pentagon's just going to hit the same wall. And if they don't, they'll look like hypocrites.

Altman says he wants to "help de-escalate" tensions. That sounds like negotiations-speak for "we want the contract, but we need a face-saving compromise."

OpenAI's ChatGPT is already available in unclassified military systems. Talks to move it into classified networks have accelerated since the Anthropic fallout.

Why This Matters

This is the first time a major AI company has been banned from government work specifically for maintaining ethical guardrails.

Not for foreign ties. Not for security breaches. For refusing to build surveillance and weapons systems without restrictions.

The precedent is set: if you build AI safety guardrails the Pentagon doesn't like, you're a national security risk.

Compare this to the Pentagon's embrace of Grok, Elon Musk's AI chatbot backed by Saudi and Qatari investors, with documented safety control failures. That model gets classified network access. Anthropic gets blacklisted.

The message is clear: weak guardrails are fine. Strong ones make you an enemy.

What Happens Next

Anthropic isn't giving up. Despite the designation, a company spokesperson said they're "continuing to engage in good faith with the Department."

Congressional Democrats will likely hold hearings. Republicans control both chambers, so oversight will face obstacles.

For the surveillance watchers: this fight isn't over AI capabilities. It's about whether the U.S. military gets to use the most advanced AI for mass domestic surveillance, and whether tech companies have any power to say no.

One company just tried. It cost them $200 million and their federal business.

Sources

Published: February 28, 2026