Federal Reserve Vice Chair for Supervision Michelle Bowman told banks that artificial intelligence is both a powerful defense against cyberattacks and a dangerous tool for malicious actors. Speaking at a Financial Stability Oversight Council roundtable on May 1, Bowman cited Anthropic’s Mythos AI model as an example of technology that can identify vulnerabilities in banking infrastructure but also be repurposed to exploit them. Her remarks followed April 2026 meetings among Treasury Secretary Scott Bessent, Fed Chair Jerome Powell and the chief executives of JPMorgan Chase, Citigroup, Bank of America and Goldman Sachs on Mythos-related cybersecurity risks. Bowman said existing model risk management guidance predates the widespread use of generative and agentic AI. She also emphasized the resource gap facing smaller banks and pointed to support through the Federal Financial Institutions Examination Council and coordination with Mastercard and Visa.