Banks catching AI agents in customer accounts and security isn’t ready

Major U.S. banks are encountering an increasing volume of third-party AI agents accessing customer accounts, often utilizing credentials voluntarily provided by users. Because traditional fraud detection models are designed to evaluate human behavior, these autonomous agents risk being incorrectly flagged as abnormal traffic or bypassing security controls entirely without proper authorization oversight. Industry experts argue that financial institutions will need to establish distinct digital identity and authorization frameworks to securely manage nonhuman actors. Open article →