Deeper Insights | AI-Powered SEO & Business Growth Solutions
President Donald Trump told all U.S. federal agencies to stop using Anthropic’s AI technology right away on February 27, 2026. This included the popular Claude models. Defence Secretary Pete Hegseth then called Anthropic a “supply-chain risk to national security,” a term that has never been used to describe a U.S. company before. This action prevents defence contractors from working with Anthropic and effectively puts the company on a blacklist for most government work.
The action is not a total ban on private users across the country, but it is the strongest response from the U.S. government so far against a domestic AI developer.
What Triggered the Clash?
The fight was about a Pentagon contract worth up to $200 million. The Department of Defence wanted “all lawful uses” of Claude, with no exceptions. Anthropic wouldn’t take out two important safety features from its Constitutional AI framework:
- No mass surveillance of American citizens at home.
- No weapons that can choose and attack targets on their own without any real human oversight.
Dario Amodei, the CEO, said that the models are not yet reliable enough for these high-risk uses. The government acted within hours of Anthropic sticking to its guns after the 5 p.m. deadline.
Key Timeline (February 2026)
- February 24–26: The Pentagon gives Anthropic an ultimatum, and Anthropic publicly defends its red lines.
- February 27: Trump’s announcement on Truth Social; Hegseth’s designation of supply-chain risk.
- On the same day, OpenAI says it got a similar deal with the Pentagon while keeping the same safety measures in place.
OpenAI Got the Deal
OpenAI CEO Sam Altman confirmed that the company had won the Pentagon contract for classified networks just hours after Trump’s announcement. OpenAI kept the same basic rules: no mass surveillance in the US and no fully autonomous weapons. They also added more layers of protection. The quick agreement allowed OpenAI to quickly fill the gap left by Anthropic.
Why the Strong Response?
The Trump administration made the issue about national security and independence. During an AI arms race with China, officials said that private companies should not be able to tell the U.S. military what to do. Trump said that Anthropic was “out-of-control Radical Left” and that it was trying to “strong-arm” the Pentagon.
Anthropic said that the designation is “legally unsound” and “unprecedented,” which is a bad example for any U.S. company that wants to do business with the government. The business said it would fight the move in court.
— Rapid Response 47 (@RapidResponse47) February 27, 2026
What It Means
- For Anthropic, there is a chance that the government will lose money and that defence contractors will have to drop Claude. Private and business use is still fine.
- For the AI field: Signals that going against military orders can lead to quick punishment. OpenAI has already made progress.
- For the safety of the country: Faster adoption of AI without private vetoes, but there is a risk that trust between Silicon Valley and the Pentagon will break down.
The legal challenge is still going on as of March 2, 2026, and the six-month phase-out for current federal use is still going on. The episode shows how the government’s demands for unrestricted military use are making AI safety principles more and more difficult to follow.
This “Trump Anthropic ban” is an important early test of how the U.S. will handle the needs of innovation, ethics, and defence in the age of AI.