White House Reconsiders Anthropic's Claude Mythos for Federal Cybersecurity Push

In a notable shift in stance, the White House is reportedly exploring the use of Anthropic's most advanced AI system, Claude Mythos, across federal agencies to strengthen cybersecurity defenses. The move comes months after tensions between the US government and Anthropic led to restrictions on the company's technology.

Earlier this year, a dispute arose when Anthropic declined to allow unrestricted military use of its AI systems. In response, the US Department of Defense designated the company as a "supply chain risk" and began transitioning toward alternatives, including offerings from competitors. At the time, Donald Trump directed federal agencies to halt the use of Anthropic's tools, stating, "We don't need it, we don't want it, and will not do business with them again!"

Despite that strong position, recent developments suggest a reconsideration may be underway. According to reports, officials within the White House believe Claude Mythos could play a critical role in identifying vulnerabilities across government systems. The AI model, which Anthropic has not released publicly due to its advanced capabilities, is said to excel at uncovering complex cybersecurity flaws.

The initiative appears to be gaining traction among senior officials. Gregory Barbaccia, the federal chief information officer at the Office of Management and Budget, recently informed Cabinet departments that a framework is being developed to enable agency-level access to Mythos. While specific agencies have not yet been named, further guidance is expected in the coming weeks.

Interest in the system is reportedly growing across government departments. Multiple chief information officers have requested access, viewing the technology as a potential game-changer in defending critical infrastructure. The goal is to leverage Mythos to proactively detect weaknesses in software systems before they can be exploited.

However, the move raises questions about whether the administration will override the Pentagon's earlier "supply chain risk" designation. Anthropic is currently challenging that label in court, arguing that it could cost the company billions in lost business. Defense Secretary Pete Hegseth had previously backed the designation, making any reversal a significant policy shift.

Interestingly, despite the official stance, reports indicate that the US military has continued limited use of Anthropic's AI tools in certain operations, including those related to Iran. These uses are expected to be phased out in the coming months as part of a broader transition strategy.

Anthropic has also taken steps to carefully control access to Claude Mythos. Through its "Project Glasswing" initiative, the company has granted limited availability to select partners, including major tech firms, to test and address vulnerabilities identified by the system.

Ahead of its release, senior US officials were briefed on both the defensive and offensive capabilities of Mythos, with input from key cybersecurity bodies. On the same day the model was unveiled, Scott Bessent and Jerome Powell reportedly encouraged financial leaders to adopt the tool for internal risk assessments.

As cybersecurity threats continue to evolve, the White House appears willing to reassess earlier decisions if it means gaining a technological edge -- potentially bringing Anthropic back into the federal AI ecosystem.