Agentic payments and the liability gap regulators haven't closed, with Kelvin Chen, Consumer Bankers Association

Kelvin Chen built the Federal Reserve's Innovation Policy Team, setting the initial frameworks for AI in supervised institutions before most regulators had settled on the right questions to ask. He held senior regulatory roles at Capital One and Barclays before becoming Head of Policy in the US at the Consumer Bankers Association.In this episode of Banking Reinvented, Kelvin joins host Tim Rutten to unpack the regulatory blind spot sitting at the center of agentic payments. The Electronic Fund Transfer Act, the law that still governs how consumers are protected from unauthorized transactions, was written in 1978 for a world where a human authorizes every transfer. That world is ending, and the liability rules for what happens when an AI agent moves money on a customer's behalf haven't been written yet.The conversation covers why banks will get pulled into agentic commerce whether they choose to or not, the regulatory vacuum left behind after recent model risk management guidance was updated without addressing generative AI, and why stablecoin rails create a new category of consumer risk with no chargeback protection.Kelvin also walks through the framework the Consumer Bankers Association is building to help close that gap, and why he believes ongoing monitoring, not upfront explainability, is the more realistic path to governing AI agents that behave probabilistically.Curious to find out more about AI in banking? Check out our content hub.