The U.S. National Security Agency (NSA) is utilizing the controversial Mythos model from Anthropic, despite the startup's ongoing conflict with the Pentagon. This was reported by Axios citing sources.
Previously, disagreements arose between Anthropic and the Pentagon due to the company's refusal to allow its AI models to be used for mass surveillance of U.S. citizens and the development of autonomous weapons. As a result, President Donald Trump ordered all federal agencies to completely cease using the startup's AI technologies within six months.
In April, Anthropic created a new model, Claude Mythos, but opted not to release it publicly due to significant security risks. During several weeks of testing, the neural network identified thousands of zero-day vulnerabilities in major operating systems and web browsers.
Instead of a public release, the company launched Project Glasswing—an initiative involving AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorgan Chase, the Linux Foundation, Microsoft, Nvidia, Palo Alto Networks, and others to test the tool in secure environments.
Axios has learned that the NSA is among the undisclosed recipients of this technology. It remains unclear how the agency is using the startup's AI, but others are employing it to scan their own environments for vulnerabilities.
“Currently, the military is expanding its use of Anthropic's tools while simultaneously claiming in court that their use poses a threat to U.S. national security,” the journalists noted.
It seems that Anthropic's relationship with the authorities has warmed. On April 18, CEO Dario Amodei discussed potential collaboration with representatives from the administration.
The meeting included White House Chief of Staff Suzy Wiles and Treasury Secretary Scott Bessent.
Recall that in April, Anthropic released an environment for launching complex and lengthy agent tasks—Claude Managed Agents.