Anthropic CEO Dario Amodei is preparing for a pivotal meeting this Friday with White House Chief of Staff Susie Wiles. The objective is to end a period of friction with the defense establishment and remove the company from a de facto "blacklist" it entered following a lawsuit against the U.S. Department of Defense. According to reports from Axios and The Wall Street Journal, the rift was initially caused by Amodei’s refusal to grant the Pentagon unrestricted access to his AI models. Now, Anthropic is pivoting, actively recruiting advisors with ties to the Trump administration to smooth over relations. While Defense Secretary Pete Hegseth labeled the company a security threat as recently as February, the upcoming meeting is intended to lay the groundwork for a major partnership.
At the heart of these negotiations is Mythos—a new project built on the Claude architecture, specifically designed to bypass and test sophisticated cyber defense systems. Current reports indicate that there will be no public release for the model; instead, tech firms and critical infrastructure operators have already been granted access to test versions. Amodei’s pitch is pragmatically framed: according to an Axios source, he argues that a U.S. failure to adopt Mythos would be a "gift to China." While Hegseth remains skeptical, the Office of Management and Budget is already preparing the necessary framework to provide federal agencies with access to the model.
Public sector interest in Mythos continues to surge. The U.S. intelligence community and the Cybersecurity and Infrastructure Security Agency (CISA) are already testing the model, and the Treasury Department has expressed focused interest. The scope of these negotiations extends beyond American borders—the European Union is also in discussions regarding the use of Mythos. The underlying logic is straightforward: the technology is perceived as so effective that for a government to reject it would border on criminal negligence.
The current situation highlights how rapidly the boundaries between private AI research and national defense are dissolving. If Anthropic successfully secures a deal, it will set a significant precedent: AI companies may increasingly use geopolitical leverage to win government contracts, even in the face of stiff resistance from traditional defense and security leadership.