The challenge for AI agents is deciding who pays for automated errors

Imagine asking an AI assistant to book a decent hotel for a weekend away. You give it a budget and say you’d like somewhere near the center, then it finds a room, pays, and sends you the confirmation.

The hotel is within budget and in a solid location, but the room has no window, breakfast costs extra, and cancellation is impossible.

The software may have followed its instructions closely enough to leave you arguing about a purchase you would never have made yourself. As payment companies build services for AI agents, that gap between permission and satisfaction becomes part of the product they have to sell.

In a report released Sept. 8, Mastercard forecasts that one in ten people will routinely use AI agents to shop and pay by 2030.

Shopping online can mean comparing a dozen tabs, decoding fees, and wondering whether the product description has omitted something essential. Software that does the tiresome part deserves a chance, but once it can spend money, its mistakes become purchases someone has to undo.

… Continue reading the full article at the original source below.