It’s certainly nice to imagine a digital assistant that can take over the weekly grocery trip, but not if that assistant is going to steal the shopper’s identity and drain their bank account.
New research from the latest edition of PYMNTS Intelligence’s exclusive Agentic AI Report shows that Americans are increasingly comfortable letting artificial intelligence (AI) take over their carts. More than half of U.S. consumers already use AI in the buying process, making it the most popular use case for the technology today.
It’s not just about asking AI what to buy anymore. Many consumers are open to letting autonomous agents handle the entire transaction. Nearly half say they’d consider letting an AI assistant shop for groceries, and 44% would trust one to pick out gifts, according to previous PYMNTS Intelligence research. The appeal is obvious: offloading time-consuming decisions and errands to software that never gets tired.
That convenience can be a slippery slope.
“Consumers will be lured by the simplicity of having a machine shop for them or do difficult planning tasks like travel and big-ticket purchases,” Ron Zayas, CEO of Ironwall by Incogni, said in an interview. “Each step along the process will seem like a logical next step: ‘as long as I am researching this for you, would you like me to purchase it for you?’”
Agentic commerce has come a long way in the last year. In January, Google launched its Universal Commerce Protocol, an open-source standard designed by the giant tech company and industry partners, to enable AI-driven, agentic commerce. It has now introduced agentic commerce through Etsy and Wayfair, with plans to add Shopify, Target and Walmart.
Advertisement: Scroll to Continue
However, major concerns still linger. Agentic commerce involves handing over a lot of information to the AI platforms and payment companies supporting those transactions. For agents to be able to make purchases, the Visas and American Expresses of the world need to know a lot about the people they’re acting on behalf of. Shoppers have their doubts. Almost all (95%) consumers have at least one concern about agentic commerce. Those worries range from simple mistakes like buying the wrong item to higher-stakes issues like identity theft.
Those anxieties aren’t coming out of nowhere. People had enough security concerns even before agentic AI was introduced. In fact, PYMNTS Intelligence finds that nearly one in five consumers have been scammed in the last five years. That figure is even higher for the most digitally connected generations: 22% of Generation Z and 24% of millennials. Now, agents come with a whole new set of risks.
“Many AI systems rely on multiple APIs and vendors,” David B. Hoppe, the founder and managing partner at technology-focused Gamma Law, told PYMNTS. “Each integration increases vulnerabilities and complicates accountability for breaches or misuse … Even ‘anonymized’ data becomes highly identifiable when combined across sources, enabling granular profiling that may [lead to] unfair targeting concerns.”
Consumers and merchants alike could grant agents unprecedented access. And if that access goes unchecked, the possibilities are concerning.
Still, many people want to trust agentic commerce, so they’re willing to negotiate. The technology promises to take the work of shopping off people’s hands, an appealing proposition for anyone who’s had to navigate their way through a department store during the holiday season or spend precious minutes of their lives deciding between nearly identical jars of pasta sauce. Half of U.S. consumers say they’d trust agentic commerce if they knew that there were fraud protections in place.
“Consumers are pragmatic and will trade data for a time-back ROI, but only if they feel in control,” Albert Roux, executive vice president of product and identity at Microblink, told PYMNTS.
Payment giants are paying close attention to those trust concerns. Take Visa, for instance. The company recently helped design Stripe and Tempo’s Machine Payments Protocol (MPP) and extended that protocol to card payments on its global network and Visa Acceptance Platform in an effort to provide secure autonomous agent payments. Rubail Birwadker, Visa’s senior vice president and global head of growth, recently explained in an interview with PYMNTS CEO Karen Webster that the technology needs “continuous validation of agent behavior” from humans to get people to trust it.
Similarly, American Express chairman and CEO Stephen J. Squeri wrote in a March letter to shareholders that having the right security measures in place is “of paramount importance” for agentic commerce to succeed. Mastercard, for its part, has partnered with PayPal to provide secure, verified agentic AI payment options.
The goal is to ensure that even if a machine is making the purchase, there’s still oversight, verification and accountability behind the scenes. The stakes are high. People are just starting to form their opinions about agentic commerce, and trust, once broken, is hard to rebuild.
“If AI is just a more sophisticated way to target and sell, that will erode trust quickly,” Trulioo CTO Hal Lonas explained to PYMNTS. “The real opportunity, and the line consumers will draw, is whether the agent is actually solving problems for them or just creating new ones.”
At PYMNTS Intelligence, we work with businesses to uncover insights that fuel intelligent, data-driven discussions on changing customer expectations, a more connected economy and the strategic shifts necessary to achieve outcomes. With rigorous research methodologies and unwavering commitment to objective quality, we offer trusted data to grow your business. As our partner, you’ll have access to our diverse team of PhDs, researchers, data analysts, number crunchers, subject matter veterans and editorial experts.
