The National Security Agency (NSA) is reportedly utilizing Mythos Preview, an advanced artificial intelligence model developed by Anthropic, despite its parent agency, the Department of Defense (Pentagon), having previously labeled Anthropic a “supply-chain risk.” This acquisition by the state's security apparatus comes after Anthropic refused to grant Pentagon officials unrestricted access to its models for purposes of mass domestic surveillance and autonomous weapons development.
Anthropic announced Mythos earlier this month as a frontier model specifically designed for cybersecurity tasks.
The company, however, stated that the model possessed capabilities too potent for offensive cyberattacks to be released publicly.
Consequently, Anthropic restricted access to Mythos to approximately 40 organizations, publicly naming only a dozen of these recipients.
The NSA appears to be among the undisclosed entities granted access to this powerful tool, reportedly employing Mythos primarily for scanning environments to identify exploitable vulnerabilities.
The U.K.’s AI Security Institute has also confirmed its access to the Mythos model, indicating a broader pattern of state security agencies acquiring advanced AI capabilities.
The State's Coercive Apparatus
The U.S. military's expanding reliance on Anthropic’s tools unfolds concurrently with its legal arguments asserting that these very tools could pose a threat to national security.
This contradiction stems from the Pentagon’s initial dispute with Anthropic, which arose when the company declined to make its Claude model available for widespread domestic surveillance and the development of autonomous weapons.
The state's legal posturing against the potential dangers of these technologies is thus revealed as a strategic maneuver to gain control over them, rather than a genuine concern for public safety.
Capital's Complicity and Negotiation
The NSA’s recent access to Mythos coincides with what appears to be a thawing relationship between Anthropic and the Trump administration.
Last Friday, Anthropic chief executive Dario Amodei held a meeting with White House chief of staff Susie Wiles and Secretary of the Treasury Scott Bessent.
The White House characterized this meeting as productive, suggesting a resolution to the previous tensions and a pathway for the state to secure access to critical technologies.
The company's refusal to comment on the matter, alongside the NSA's silence when TechCrunch reached out for comment, underscores the opaque nature of these negotiations between private capital and state power.