The Trump administration placed Anthropic on a blacklist, but now it is encouraging banks to utilize its AI technology.

      In summary, Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell are encouraging Wall Street's largest banks to evaluate Anthropic's Mythos AI model for cybersecurity issues, despite the Pentagon's legal battle against Anthropic after labeling it a supply chain risk due to its refusal to eliminate safety measures on autonomous weapons and mass surveillance. Reports indicate that JPMorgan Chase, Goldman Sachs, Citigroup, Bank of America, and Morgan Stanley are currently testing the model. Mythos has identified thousands of zero-day vulnerabilities in major operating systems and browsers and is being distributed through a limited program called Project Glasswing to around 50 organizations. UK regulators are also working to assess potential risks.

      The Trump administration is discreetly prodding the nation's biggest banks to assess the technology of the same AI firm it has been attempting to undermine for the past two months. This week, Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell called on executives from financial giants like JPMorgan Chase, Goldman Sachs, Citigroup, Bank of America, and Morgan Stanley to utilize Anthropic's new Mythos model for identifying cybersecurity vulnerabilities, as reported by Bloomberg.

      This recommendation stands out for its inherent contradiction. Anthropic is presently engaged in a legal dispute with the Department of Defense after Defense Secretary Pete Hegseth classified the company as a "supply chain risk," which disqualifies it from military contracts and compels defense contractors to cease utilizing its technology. This classification followed Anthropic's refusal to withdraw two safety measures from its AI models: banning their use in fully autonomous weapons and prohibiting mass surveillance of U.S. citizens.

      Now, two of the administration's top economic figures are urging Wall Street to embrace the very product that the Pentagon has sought to ban.

      What Mythos actually accomplishes

      Claude Mythos Preview is an advanced model that Anthropic did not specifically design for cybersecurity. Its ability to discover vulnerabilities arose as a secondary effect of general enhancements in code reasoning and autonomous functioning. During evaluations, Mythos uncovered thousands of zero-day vulnerabilities, which are flaws previously unknown to developers, across all major operating systems and web browsers.

      The findings were significant enough for Anthropic to decide against a public release of the model. Instead, they initiated Project Glasswing, a controlled program providing access to approximately 50 organizations, including Amazon Web Services, Apple, Google, Microsoft, Nvidia, Cisco, CrowdStrike, Palo Alto Networks, and JPMorgan Chase. Anthropic has allocated up to $100 million in usage credits and $4 million in direct contributions to open-source security organizations as part of this initiative.

      This characterization of the model as "too dangerous to release" has raised skepticism. Tom's Hardware pointed out that assertions of "thousands" of serious zero-day discoveries were based on merely 198 manual reviews, and many of the flagged vulnerabilities were in older software or difficult to exploit. Others in the security field suggested that the limited release seemed less about responsible AI governance and more like a savvy business strategy: create scarcity, instill fear, and attract customers.

      The Pentagon paradox

      The clash between Bessent and Powell's recommendation and Hegseth's designation reflects two branches of the same administration pursuing openly contradictory policies regarding the same company. The Pentagon's dispute began in February, when Hegseth gave Anthropic CEO Dario Amodei a deadline to remove the company's safety measures or jeopardize its $200 million defense contract. Amodei declined the request. In response, Hegseth labeled Anthropic as a supply chain risk, and President Trump instructed federal agencies to refrain from using its technology. A Pentagon official accused Amodei of having a "God complex," while Trump condemned Anthropic as a "radical left, woke company."

      The courts have since reached a split decision. A federal judge in California issued a preliminary injunction to block the supply chain designation, stating that "nothing in the governing statute supports the Orwellian notion that an American company may be branded a potential adversary and saboteur of the US for expressing disagreement with the government." However, an appeals court in Washington, D.C., denied Anthropic's request to temporarily halt the blacklisting as the case progresses. Consequently, Anthropic is barred from Department of Defense contracts but can continue collaborating with other government agencies.

      It is within this context, excluded from the Pentagon yet not from the Treasury or the Fed, that Bessent and Powell intervened this week.

      What the banks are doing

      While JPMorgan Chase is officially recognized as a Project Glasswing partner, Bloomberg has reported that Goldman Sachs, Citigroup, Bank of America, and Morgan Stanley are all conducting internal tests of Mythos. The applications reportedly include vulnerability detection, fraud-risk identification, and compliance workflow automation within financial systems.

      The rapid adoption is indicative of a serious concern. If Mythos can detect zero-day vulnerabilities in operating systems and browsers, it is likely capable of finding them in banking infrastructure as well, along with any sufficiently advanced model that