Anthropic Files Declarations Refuting DoD AI Risk Claims

Artificial intelligence company Anthropic has intensified its legal challenge against the U.S. Department of Defense, filing sworn declarations in a California federal court that contradict the Pentagon's claims of national security risks posed by the firm's technology. The filings, submitted late Friday ahead of a March 24 hearing in San Francisco, argue that the government's designation of Anthropic as a supply chain risk relies on technical inaccuracies and unraised concerns from prior negotiations. This dispute, which began with a public breakdown in late February, centers on Anthropic's refusal to grant the military unrestricted access to its Claude AI model without safeguards against uses like autonomous weapons and mass surveillance of U.S. citizens.

The conflict escalated when the Pentagon formalized the supply chain risk designation on March 3, effectively barring the department and its contractors from using Claude in the coming months. Anthropic filed its lawsuit on March 9, seeking a temporary restraining order to halt the ban. Court documents reveal that negotiations had appeared productive until shortly before the designation. On February 24, Anthropic CEO Dario...