US' NSA Using Anthropic's Mythos AI Model Despite Pentagon's 'Supply Chain Risk' Designation

The National Security Agency (NSA) is using Anthropic's most powerful AI model to date, Mythos Preview, even as top officials at the Department of Defense, which oversees the NSA, have labeled the company a 'supply chain risk,' according to sources familiar with the matter.

The development highlights tensions within the US government over the use of advanced AI tools for cybersecurity, with national security needs appearing to clash with ongoing disputes involving the Pentagon and Anthropic.

NSA is reportedly using Mythos Preview

Two sources told Axios that the NSA has access to Mythos Preview, while a third source indicated the model is also being used more widely within the Department of Defense. It remains unclear exactly how the NSA is deploying the model, though other organisations with access primarily use it to scan their environments for exploitable security vulnerabilities.

Anthropic has restricted access to Mythos Preview to approximately 40 organisations, citing the model's strong offensive cyber capabilities as too dangerous for broader release. The company publicly announced only 12 of those organisations; one source said the NSA was among the unnamed agencies granted access. The UK's counterparts to the NSA have also confirmed access to the model through the country's AI Security Institute.

The situation stems from a broader feud between Anthropic and the Pentagon. In February, the Defense Department moved to cut off Anthropic and directed its vendors to do the same, with that case still ongoing. The military has argued in court that using the company's tools could threaten US national security, even as it broadens internal use of Anthropic's technology.

The rift intensified during contract renegotiations earlier this year. The Pentagon reportedly demanded that Anthropic make its Claude model available for 'all lawful purposes,' while the company pushed to restrict uses involving mass domestic surveillance and the development of autonomous weapons. Some Defense officials have argued that Anthropic's stance shows it cannot be trusted in critical military situations, a claim the company denies. Others within the administration have expressed a desire to resolve the dispute to access Anthropic's cutting-edge tools.

Amodei met with White House chief

On Friday, Anthropic CEO Dario Amodei met with White House Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent to discuss government use of Mythos and the company's broader plans and security practices. Sources described the meeting as productive, with next steps expected to focus on engagement by departments outside the Pentagon.