NSA Using Anthropic's Unreleased Mythos Model While Pentagon Labels Company Supply Chain Risk

The National Security Agency is using Anthropic's Mythos Preview model for vulnerability scanning, according to Axios, even as the agency's parent organization labels the AI company a national security threat.

The NSA is among approximately 40 organizations with access to Mythos Preview, Anthropic's frontier cybersecurity model announced earlier this month. Anthropic withheld the model from public release, claiming it is too capable of offensive cyberattacks. The company has publicly named only a dozen of the organizations with access. The UK's AI Security Institute has confirmed it also has Mythos access.

The NSA is using Mythos primarily for scanning environments for exploitable vulnerabilities, according to the report.

Pentagon Dispute

The deployment creates a contradiction within the U.S. defense establishment. Weeks before the NSA's Mythos usage was reported, the Department of Defense labeled Anthropic a "supply chain risk." The designation followed Anthropic's refusal to allow Pentagon officials unrestricted access to its models' full capabilities.

The Pentagon's dispute with Anthropic originated when the company refused to make Claude available for mass domestic surveillance and autonomous weapons development. The U.S. military is now simultaneously using Anthropic's tools while arguing in court that those tools threaten national security.

Thawing Relations

Anthropic's relationship with the Trump administration appears to be improving. On April 17, Anthropic CEO Dario Amodei met with White House Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent. The White House characterized the meeting as productive.

Anthropic declined to comment on the NSA's use of Mythos. The NSA has not responded to requests for comment.

What This Means

The NSA's Mythos deployment reveals the pragmatic reality beneath bureaucratic disputes: intelligence agencies will use effective tools regardless of broader policy conflicts. The situation also highlights the challenge AI companies face when government customers simultaneously demand access while raising security concerns—a dynamic that will likely intensify as models grow more capable. Anthropic's selective distribution strategy for Mythos creates a two-tier system where trusted organizations get offensive-capable tools while the broader research community cannot audit their risks.