On Friday, the Pentagon's tech chief made it clear: Anthropic's Claude models are still a no-go for the Department of Defense. Emil Michael, the DOD's technology leader, told CNBC's "Squawk Box" that Claude remains a supply chain risk, and other government departments may phase it out over the coming months. But he drew a sharp distinction between Claude and another AI model: Mythos.

Michael called Mythos a "separate national security moment." Why? Because Mythos has advanced cyber capabilities that could actually identify and fix vulnerabilities in networks. That's a double-edged sword—it can help secure systems, but it also poses risks if it falls into the wrong hands. Michael said some "exceptions" might be granted for Claude based on integration complexities, but Mythos is a whole different ballgame. The issue isn't just with the DOD, he stressed; it's a government-wide concern that requires everyone to fortify their networks.

Pentagon Signs AI Deals With Major Firms

Also on Friday, the DOD announced it had signed agreements with seven companies to deploy their AI technology on the agency's classified networks for "lawful operational use." The list reads like a who's who of tech: Alphabet (Alphabet (GOOGL) and GOOG), OpenAI, Nvidia (NVDA), Microsoft (MSFT), Amazon (AMZN), Elon Musk's SpaceX, and a startup called Reflection. This is part of a push to diversify AI providers and reduce reliance on any single model.

Michael said the Pentagon still wants guardrails on AI use, but the terms are "negotiable" depending on the company. Different firms have different views, so the DOD is keeping things flexible.

White House AI Push After Anthropic Talks

This all comes after reports that the Trump administration was trying to find a workaround to Anthropic's supply chain risk designation, specifically to deploy Mythos for government use. The White House was reportedly drafting an executive action to bypass the ban. That move followed a meeting between White House Chief of Staff Susie Wiles, Treasury Secretary Scott Bessent, and Anthropic CEO Dario Amodei, which was described as a "productive introductory meeting" on potential collaborations.

But tensions with Anthropic aside, White House officials are also drafting comprehensive AI guidelines. The policy encourages U.S. agencies to diversify their AI providers to avoid dependence on a single model. It also stipulates that AI companies contracting with the Department of Defense must not interfere in the military's chain of command. So while the door might be open for Mythos, the rules are getting tighter for everyone.