The head of Elavon, a payments company owned by U.S. Bank, says the push to let AI assistants shop for people needs to slow down until safety is sorted out. CEO Wally Mlynarski spoke at the Money20/20 conference in Las Vegas last week. His warning was blunt: if the industry races in and does it wrong, "consumers are going to be harmed."
Why he is worried
Agentic commerce means AI agents that make purchases on your behalf. Mlynarski said the industry has spent 25 years chasing online fraud, "and we've made all the mistakes." He pointed to card security codes, industry data security standards and keyed-in card entries as tools built to manage risk. "We could go screw that up again with agentic commerce," he said.
Who pays when it goes wrong
Big-bank leaders, including JPMorgan Chase's Jamie Dimon and PNC's Bill Demchak, have raised liability concerns about AI agents acting for customers. Liability means who is responsible for the loss when something goes wrong. The article does not say how that question will be settled.
Mlynarski also said the money side will be hard. Adding agents to a chain that already includes the shopper, the merchant, the merchant's bank, the card issuer and the card networks means someone's cut may have to change.
What is already happening
Meta has added Stripe's Link digital wallet so its personal AI agent, Muse, can make payments. Consumers have also started using Instinct and OpenAI's dots. Mlynarski said such launches and pilots are "not necessarily thinking about all the safety and soundness in there," and contrasted them with efforts led by banks.
The bank-backed option
His answer is Paze, a mobile wallet run by Early Warning Services, which is owned by seven large U.S. banks, including U.S. Bank. He said Paze is "ready to do pilots." He said it can work within agentic commerce and address agent hallucination, which is when an AI makes things up or gets facts wrong. He helped shape Paze during his time at Bank of America. He is also speaking for a bank-owned product, so his view is not neutral.
What is not known yet
No pilot dates were given. Standards and protocols for AI shopping are still being worked out. There is no clear answer on who covers the loss if an agent buys the wrong thing. Until that is settled, it is wise to be careful about what an AI assistant is allowed to buy for you.
The takeawayIf an AI assistant offers to buy things for you, remember that the rules on who covers its mistakes are still unsettled.