NSA Using Claude Mythos Preview for Vulnerability Discovery
The NSA has continued deploying Claude Mythos Preview for vulnerability discovery even as Anthropic carries a supply-chain risk status — with federal agencies directed to stop using its technology — and the Pentagon remains in an active legal dispute with the company, per Axios reporting. The NSA's continued use illustrates a core dual-use tension: the same models flagged as highest-risk may be most strategically valuable to national security agencies pursuing offensive and defensive cyber operations.
Why It Matters
The gap between official risk designations and operational usage signals that AI capability now overrides procurement policy in high-stakes security contexts — a pattern that will shape future AI governance frameworks.