Government & Policy
DOD calls Anthropic an “unacceptable risk to national security” over model “red lines”
A rare moment where the policy fight is spelled out in plain language: the Pentagon is arguing that Anthropic could disable or alter its model behavior if it believes its internal “red lines” are being crossed during operations. That argument sits inside a 40‑page court filing as Anthropic seeks to block a “supply chain risk” designation. The backstory matters: Anthropic reportedly pushed back on uses like mass surveillance of Americans and the use of models in targeting/firing decisions for lethal weapons. Even if this ends as a contract dispute, it’s a preview of how “AI governance” becomes a battlefield claim.
Source: TechCrunch