The intersection of advanced artificial intelligence and national defense security has become a highly complex, and often contentious, area. Recent reports, cited by Axios, reveal that the National Security Agency (NSA) has resumed or expanded its use of Anthropic’s Mythos Preview model, a development that directly clashes with the Pentagon’s official policy recommendations. The core conflict centers on the management of ‘supply chain risk,’ a term frequently used by defense departments to describe vulnerabilities introduced by third-party vendors and technologies.
The Department of Defense had previously attempted a decisive intervention, moving in February to restrict or cut off access to Anthropic’s services, attempting to compel its various corporate vendors to follow suit. This institutional struggle remains legally active. However, according to sources, the military is not only increasing its reliance on Anthropic’s tools but is doing so while concurrently challenging the legality and implications of its own actions in court. This suggests a tension: the urgent operational necessity of having cutting-edge AI capabilities versus the rigid bureaucratic and legal demands for vendor oversight.
Mythos itself is not available to the general public. Anthropic has significantly limited access to the model, restricting it to approximately forty organizations worldwide due to what the company deems its