The National Security Agency is using Anthropic’s new Mythos Preview AI model, even as the company remains locked in a bitter legal fight with the Pentagon over military applications. Sources familiar with the matter revealed that the NSA received access to the cybersecurity-focused AI tool as part of a limited preview program.
This development highlights the complex relationship between AI companies and government agencies. While Anthropic refuses to compromise on safety guardrails for military use, certain government departments are still finding ways to access its most advanced technology.
What makes Mythos different
Anthropic launched Mythos Preview in early April, marketing it as a general-purpose language model with particular strength in computer security tasks. The company describes the AI as “strikingly capable” at cybersecurity work, making it an attractive tool for intelligence agencies like the NSA.
The NSA is among roughly 40 organizations that received early access to Mythos Preview. Sources suggest the model is “being used more widely within the department,” indicating broader adoption beyond initial testing.
The ongoing government standoff
The NSA’s use of Mythos comes amid a months-long dispute between Anthropic and the Trump administration. In February, President Trump ordered all government agencies to stop using Anthropic’s services after the company refused to relax certain safeguards during military contract negotiations.
The conflict escalated when the Trump administration labeled Anthropic a “supply chain risk.” The company responded by filing lawsuits against the Department of Defense in March, challenging this designation in two separate courts.
Legal proceedings have produced mixed results so far:
- One court granted Anthropic a preliminary injunction, temporarily blocking the supply chain risk label
- Federal judges in another court denied the company’s motion to lift the designation
- The Pentagon has filed responses to defend its position
White House meetings signal potential thaw
Recent diplomatic efforts suggest both sides may be looking for a resolution. Anthropic CEO Dario Amodei met with White House chief of staff Susie Wiles and other officials on Friday, reportedly to discuss Mythos and the broader relationship between the company and government.
The White House characterized the meeting as “productive and constructive,” though President Trump claimed he had “no idea” about the discussion when questioned by reporters. This disconnect illustrates the complex dynamics within the administration regarding AI policy and vendor relationships.
The situation reflects broader tensions in the AI industry over how these powerful tools should be used by military and intelligence agencies. As AI capabilities advance, companies face increasing pressure to balance commercial opportunities with ethical concerns about weaponization and surveillance applications.