NSA Is Using Anthropic's Mythos Preview Despite Pentagon Supply Chain Risk Designation
The US National Security Agency is actively using Anthropic's Mythos Preview AI model, according to sources cited by Axios — a striking development given that Anthropic has been officially designated as a supply chain risk by the Pentagon. Sources say Mythos is also in widespread use across the Department of Defense, with roughly 30 unannounced groups piloting the system.
The Supply Chain Risk Paradox
The Pentagon's designation of Anthropic as a supply chain risk was intended to limit the company's access to military contracts. Yet the NSA's adoption of Mythos Preview suggests that designation hasn't stopped operational units from deploying the model internally. One source noted that the NSA was part of the early rollout groups, suggesting they were likely "in from the start."
The move raises legal and policy questions: does the NSA's use of Mythos undermine the Department's upcoming legal proceedings in the Ninth Circuit related to Anthropic's supply chain risk classification?
Why Anthropic Is Holding Back Mythos
A separate Financial Times report reveals that Anthropic is deliberately holding back a wider public release of Mythos. The reason: reliability. The company has suffered multiple outages in recent weeks as usage surged, and it wants to ensure it can consistently serve customers before opening the floodgates.
This capacity constraint is a growing concern for Anthropic's enterprise ambitions. Bloomberg also reported that Mythos' rapid adoption is straining open-source security teams, with maintainers receiving a "crazy" number of AI-generated bug reports that are difficult to triage.
The Bigger Picture for AI in Government
The NSA's adoption of Mythos reflects a broader trend: US intelligence agencies are moving quickly to integrate frontier AI models regardless of bureaucratic hurdles. The speed of adoption often outpaces formal procurement and security review processes, creating grey zones where powerful AI tools are in use without full institutional oversight.
The Bottom Line
The NSA using Anthropic's Mythos despite a formal supply chain risk designation is a signal that government AI adoption is accelerating faster than policy frameworks can accommodate. It also puts pressure on Anthropic to scale its infrastructure — and on policymakers to clarify where AI governance boundaries actually lie.
Related Articles
- Google Forms AI Coding Strike Team as Sergey Brin Pushes DeepMind to Pivot to Agents
- Anthropic Valued at Over $1 Trillion