A Better Newspaper

## NSA Anthropic Mythos Deployment – Policy Reversal ### Overview The NSA is reportedly using Anthropic's Mythos AI model despite an existing Pentagon blacklist barring its use (Axios, April 19, 2026). This development represents a significant fracture in US government AI procurement policy and has direct implications for the ongoing litigation surrounding Anthropic's Pentagon blacklisting. ### Background Mythos is Anthropic's restricted cybersecurity-focused AI model. A separate wiki page documents the Anthropic Pentagon Blacklisting and the subsequent Appeals Court Emergency Stay Denial (April 2026). The blacklist was understood to apply across defense and intelligence community agencies. ### Key Developments - The NSA is reportedly using Mythos despite its placement on a Pentagon blacklist (Axios, April 19, 2026) - This reportedly occurred notwithstanding the blacklist's existence, suggesting either a formal carve-out, informal workaround, or interagency policy disagreement - The Bank of Canada and Bank of England have separately convened emergency reviews related to Mythos (existing wiki pages) ### Legal & Policy Implications - The NSA's reported use may undermine the legal rationale for the Pentagon blacklist in ongoing litigation - Interagency inconsistency in AI procurement creates arguments for Anthropic in its blacklisting appeals - Congressional oversight implications: use of a blacklisted AI tool by an intelligence agency raises accountability questions - Procurement compliance risk for defense contractors who may have self-restricted Mythos use in reliance on the blacklist ### Strategic Watch This story is likely to develop further as congressional oversight bodies, the Pentagon, and Anthropic's legal team all have incentives to respond. The fracture between NSA practice and Pentagon policy may signal a broader interagency AI governance dispute.