AI Could Soon Shop for You. Can We Trust It With Our Credit Cards?

Visa announced Wednesday it is partnering with leading AI chatbot developers like Anthropic, Microsoft, OpenAI and Perplexity to connect their AI systems to Visa’s payments network and take online shopping out of consumer hands.

“Magical” AI technologies have the potential to “radically transform commerce,” Visa CEO Ryan McInerney said at this week’s Visa Product Drop 2025. Visa is also working with IBM, online payment company Stripe and phone-maker Samsung on the initiative.

Why do consumers need Visa et al. to insert themselves in the process? Trust and safety — at least, according to the company. Because, as it is now, within milliseconds of an attempted purchase, Visa is able to sift through vast amounts of data to assess whether patterns imply there’s fraud or a sale going on.

“Visa recognizes these patterns through our data. We understand the rhythms, and those rhythms are what enable us to build solutions that deliver better buyer experiences, that grow sales for sellers with less fraud, and less declines and more completed transactions,” McInerney explained.

Leaving aside that last time you spotted fraud on your bill that your credit card issuer and Visa didn’t, there will be kinks to work out, said Dr. Jennifer King, a privacy and data policy fellow at the Stanford Center for Human Centered Artificial Intelligence.

“It’s possible the first places where these [AI agents] work are trusted partners. So if I want my Starbucks latte to be ordered for me and ready every morning on my walk to work, I just pick it up. I know what it costs, and I’ll get the receipt from Starbucks eventually through my system. There’s no negotiation there. It’s just straightforward,” King said.

With more complex or one-time purchases, it’s less clear whether AI shop bots can guarantee the consumer is getting a desirable deal. That’s because we’re all shopping in what’s called the ad surveillance economy, where thousands of companies you’ve never heard of buy and sell all sorts of personal data. That impacts the prices you’re offered through the growing practice of dynamic pricing, where retailers sift through all that data to determine the highest price any individual consumer is likely to pay for a product and service.

“This ends up being a world potentially where consumers are captured even more than they are now, and are subject to higher prices and potentially deceptive conduct that they aren’t even seeing, because they’re no longer the one making the decision,” King said.

“There are so many sticky situations you can get into,” added David Harris, a senior policy advisor of the California Initiative for Technology and Democracy. Harris also lectures on AI ethics and social media at UC Berkeley’s Haas School of Business and previously worked as a research manager on the responsible AI team at Meta.

Harris recalled the court case where Air Canada was ordered to pay compensation after its AI chatbot misled a customer into believing he could get the difference between a full-price ticket and a discounted bereavement fare refunded. The airline tried to argue the bot was a “separate legal entity” that was “responsible for its own actions.”

Given the paucity of regulatory oversight and the slow and uneven response of courts, what hope do any of us have in this brave new world rolling out in the near future? Competition, said Dr. King.

“All of this is going to require trust. If you can’t trust these [AI] agents to operate on your behalf, people aren’t going to adopt them,” Dr. King said. “There’s been such a trust deficit in online services. This, to my mind, ups the ante. If people are finding out that they’re getting ripped off left and right by using these tools, they’re not going to use them.”

According to Visa, its pilot projects began this week, ahead of more widespread usage expected next year.