A group of 200-plus merchants is working with a payments body on rules for AI shopping agents, including which card they use and who is liable when a purchase goes wrong.
Two industry groups have teamed up to set ground rules for AI agents that shop for people. The Secure Technology Alliance and the Merchant Advisory Group (MAG) say they want the payments industry to agree on how these agents start, verify and complete a purchase for a consumer or a business.
The groups named three focus areas. One is cutting false declines, which are cases where a legitimate card payment is wrongly rejected. Another is defining what data stores must share so an AI agent can buy from them. The third is how a payment card gets chosen when an agent makes the purchase.
They also point to an open question: where responsibility sits when something goes wrong. An agent may follow an earlier instruction or guess what you want in the moment, using checkout systems built for people. The material does not say how that will be settled.
MAG represents more than 200 merchants, including airlines, grocers, restaurants, insurers and software firms, in North America, Europe and Asia. The Alliance is also inviting AI model developers and fraud prevention companies to join. The stated aim is to set shared practices early, before competing approaches split the market.
No rules, timeline or finished standards have been announced. Trust is also a hurdle. A SEON survey found half of surveyed US adults are uncomfortable with AI agents using their card details.
AI shopping agents are coming, but the rules on card choice and liability for mistakes are still being written, so go slowly before handing one your card.
Mastercard describes AI tools that could buy things for you, and says trust is the hard part: who gets to spend your money, and under what rules.
Read more: Mastercard outlines how AI assistants could shop and pay for youWorld models, AI that learns physics from video, are being built into home robots and driverless cars, and your own videos may help train them.
Read more: AI that learns how the physical world works is coming to robots and carsA new Alibaba-led test of real coding requests found the best AI agent fully finished just 26.4 percent of tasks, so human review still matters.
Read more: AI coding helpers fully finish only about 1 in 4 real tasks
