The Hypocrisy of National Security AI
The National Security Agency is reportedly running Anthropic’s Mythos Preview, a model the company deemed too dangerous for public release. According to Axios, the spy agency is using the tool to scan for exploitable vulnerabilities. This is the same model Anthropic kept from the open market, claiming its offensive cyber capabilities were too potent. The NSA’s parent organization, the Department of Defense, recently labeled Anthropic a “supply-chain risk” after the company refused to let Pentagon brass run unrestricted tests on Claude for mass surveillance and autonomous weapons. So the Pentagon’s own intelligence arm is now cozying up to the very technology its leadership publicly vilifies.
A Thawing Relationship and a Troubling Precedent
The NSA’s quiet adoption of Mythos coincides with a notable political detente. Last Friday, Anthropic CEO Dario Amodei met with White House Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent. The White House called the meeting productive. This suggests that the earlier feud was less about genuine security concerns and more about leverage. Anthropic wanted to control its technology’s use in warfare. The Pentagon wanted unfettered access. Now, the NSA gets access through a back channel, and Anthropic gets a seat at the table. It is a cynical trade. The lesson is clear: if you build the most powerful cyber weapon, the government will find a way to use it, regardless of your stated ethics. This undermines Anthropic’s safety theater and exposes the Pentagon’s strategic posturing.
Source: Techcrunch
