NSA Uses Anthropic's Restricted Mythos AI Model Despite Pentagon Conflict
The National Security Agency is leveraging Anthropic's Mythos Preview, an AI cybersecurity model that Anthropic chose not to release publicly due to worries about its potential for misuse. This development comes after the Department of Defense identified Anthropic as a supply chain risk when the company refused to grant unrestricted access to its models. Mythos, announced earlier this month, is restricted to approximately 40 organizations, including the NSA, which is using it mainly for vulnerability scanning. The UK's AI Security Institute has also confirmed its access to Mythos. Meanwhile, the U.S. military is increasing its reliance on Anthropic's tools while asserting in court that they pose a national security risk. Anthropic's rapport with the Trump administration improved following a constructive meeting last Friday.
Key facts
- The NSA is reportedly using Anthropic's Mythos Preview AI model.
- The Department of Defense labeled Anthropic a 'supply chain risk'.
- Anthropic announced Mythos earlier this month as a frontier model for cybersecurity.
- Mythos was withheld from public release due to offensive cyberattack capabilities.
- Access to Mythos is limited to around 40 organizations, with only about a dozen named publicly.
- The NSA is using Mythos primarily to scan for exploitable vulnerabilities.
- The UK's AI Security Institute has confirmed access to Mythos.
- Anthropic CEO Dario Amodei met with White House officials last Friday.
Entities
Artists
- Dario Amodei
- Susie Wiles
- Scott Bessent
Institutions
- National Security Agency
- Anthropic
- Department of Defense
- Pentagon
- Axios
- UK's AI Security Institute
- U.S. military
- White House
- TechCrunch
Locations
- United States
- UK