Interactive: Would you trust AI to control your money?

New AI “agents” that can act on a user’s behalf—paying bills, moving money and executing decisions—are arriving in banking, raising questions about how much control customers are willing to hand over. The article frames the core risk as autonomy: the same capability that makes an agent useful can also produce harmful outcomes if guardrails are missing. Experts interviewed describe scenarios where an agent follows instructions in an unanticipated way, not necessarily due to a hacker. Jake Williams of IANS says a “rogue agent” doesn’t require malicious intent, while T.J. Marlin of Guardrail Technologies highlights the human shift from personal control to software. The piece also cites claims that banks have spent months building AI-based defenses against rogue-agent threats.