Major US banks are blocking third-party AI agents from operating inside customer accounts. They are also stopping these agents from making transactions. The move follows reports of agents accessing accounts and acting without user instructions. Banks flag and block transactions where customers share login credentials. Morgan Stanley has blocked agents but has not seen unauthorized access. JPMorgan Chase is developing tools to manage the risk of agents acting independently. Charles Schwab is also blocking unauthorized agents. Banking executives warn that AI agents add significant risks for customers and institutions.