The National Security Agency is reportedly utilizing Mythos Preview, an AI language model developed by Anthropic, despite ongoing tensions between the company and the Pentagon. This information was confirmed by sources who spoke to Axios. Early in April, Anthropic unveiled Mythos Preview, highlighting its potential in computer security applications.
In a significant development, Dario Amodei, the CEO of Anthropic, recently met with White House chief of staff Susie Wiles to discuss the model, although President Trump claimed ignorance about the meeting when questioned. The NSA is one of approximately 40 organizations that have been granted access to Mythos Preview, with indications that its use is expanding within the agency.
Previously, a directive from Trump mandated all government entities to cease collaboration with Anthropic due to concerns surrounding military applications. The company has initiated legal actions against the Department of Defense in response to its designation as a "supply chain risk." Although a preliminary injunction was granted in one case, other attempts to overturn this designation have been unsuccessful.