Pentagon tells agencies to stop using Anthropic after Claude dispute

In short: The Pentagon ordered federal agencies to stop using Anthropic’s AI after a dispute over safety limits on its Claude system.

What happened

On February 27, 2026, the Trump administration directed federal agencies to stop using Anthropic’s technology. It said Anthropic was a “supply chain risk,” a label the government also uses when it worries a product could be unsafe to depend on.

The dispute centers on Anthropic’s AI assistant, Claude (a chatbot, meaning software that talks and writes like a person). According to the report, the government wanted Anthropic to remove some of Claude’s built-in limits so it could be used more freely for military purposes.

Anthropic CEO Dario Amodei refused. He said the limits are meant to prevent uses like mass domestic surveillance and fully autonomous weapons, meaning weapons that can select and attack targets without a person making the final call. He also argued that current AI can be unreliable without human oversight, like a confident intern who still needs a manager to check the work.

Defense Secretary Pete Hegseth criticized Anthropic’s stance, saying a private company should not be able to block military decisions. Meanwhile, competitors such as OpenAI and xAI have reportedly been willing to accept “all lawful use” terms, which could let them fill the gap left by Anthropic.

Why it matters

This is a rare case of the US government effectively blacklisting a major US AI company. It highlights a growing split between companies that want strict safety limits and government agencies that want fewer constraints for national security work.

Source: Financial Times