Fed’s Bowman warns AI could be used against banks

At a Financial Stability Oversight Council roundtable on May 1, 2026, Federal Reserve Vice Chair for Supervision Michelle Bowman warned that artificial intelligence tools moving into banking could be repurposed by attackers. Bowman pointed to Anthropic’s Mythos, a system that can scan software for vulnerabilities, and noted that capabilities used to find weaknesses could be turned against institutions.

Regulators are assessing whether existing model risk frameworks are adequate for newer generative AI systems. Traditional model rules rely on predictable behavior and explainable outcomes. Generative models can be less predictable and harder to test, monitor and explain, which complicates oversight.

Officials from the Federal Reserve, the Office of the Comptroller of the Currency and the Federal Deposit Insurance Corporation are working together on guidance that would set expectations for how banks adopt and supervise AI. The effort favors supervisory guidance over formal rulemaking for now, leaving some uncertainty about where regulatory boundaries will be drawn.

Senior officials have already engaged banks on the issue. Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell met with major banks in April to review exposures and preparedness as AI tools expand into areas such as risk management, fraud detection and customer service.

The debate on AI governance is also playing out inside government. In April the Pentagon designated Anthropic a supply-chain risk after the company declined to loosen safeguards on how its AI is used. The White House has pursued options to preserve agency access to advanced AI tools while officials assess security concerns.

International rules differ. The European Union’s AI Act imposes stricter requirements for high-risk systems, including those used in finance, while U.S. agencies are relying more on broad principles and supervisory guidance. That divergence could create different compliance expectations for banks operating across jurisdictions.

Market observers have highlighted possible capital shifts related to AI. Macro strategist Lyn Alden warned that AI-related stocks could reach a peak and prompt some investors to reallocate capital into assets such as Bitcoin. Investor Raoul Pal described both AI and crypto as driven by network effects, where value rises with wider adoption.

Bowman urged coordinated oversight across agencies to reduce fragmented approaches. Interagency work over the coming months will shape how banks are expected to test, monitor and document AI-driven models.