Anthropic and the Pentagon are clashing over limits on military AI.

The Pentagon asked Anthropic to remove internal limits on use of its AI for surveillance and autonomous weapons, and Anthropic declined, saying its systems are not reliable for fully autonomous weapons. The administration announced it would pause federal work with Anthropic and a Friday deadline had been set for the company to respond.

Anthropic and the Pentagon are engaged in a dispute over the company’s self-imposed limits on how its AI can be used. The disagreement centers on Anthropic’s restrictions on mass domestic surveillance and fully autonomous weapons and how those limits interact with military needs. Pentagon officials said they wanted the restrictions removed for certain defense uses, while Anthropic’s CEO said the company would not provide systems it considers unreliable for autonomous weapons. A Friday deadline had been set for Anthropic to respond to the Pentagon’s demands. Key points: - The Pentagon asked Anthropic to lift internal limits on use of its models for surveillance and autonomous weapons, according to officials. - Anthropic CEO Dario Amodei said frontier AI systems are not reliable enough to power fully autonomous weapons and that the company would not knowingly provide such products. - The administration announced that federal agencies would stop working with Anthropic and a Pentagon official described the company as a supply‑chain risk; other AI vendors reported reaching agreements with the Defense Department. - A Friday deadline had been set for Anthropic to agree to the Pentagon’s demands. Summary: The dispute highlights a broader debate over whether advanced AI should be treated primarily as a military tool or as a technology that developers may limit for safety reasons. Officials have paused federal work with Anthropic and set a deadline for its response. Undetermined at this time.