72% of US Consumers Have Used AI, But Only 23% Trust It With Payments

Americans Love Their AI Assistants; Just Don’t Ask Them to Pay the Bill

Most of us have already let AI into our daily routines.

Nearly three out of four US consumers, 72%, according to Visa’s new Trust Index, have used some kind of AI assistant. Ask that same group whether they’d let generative AI actually spend their money, though, and the number collapses to just 23%.

That’s not a small gap. It’s a chasm, and it says something important about where this technology actually stands with regular people.

Everyone’s using AI. Almost no one trusts it with their wallet.

The 72% adoption figure isn’t surprising if you’ve paid attention to the last couple of years. People ask AI to help them shop around, compare products, plan trips, and answer random questions; it’s become background infrastructure for a lot of daily decision-making.

Naturally, this has fueled talk of “agentic commerce,” the idea that an AI agent won’t just help you research a purchase; it’ll actually go make it. Find the product, compare a few options, pick the best one, and check out. No human hands required.

Visa’s data suggests that the last part actually pulling the trigger on a payment is where consumers draw a hard line.

Using AI and trusting AI with money aren’t the same thing.

Only 23% of people said they’d trust generative AI to handle a payment on their behalf. Compare that to 72% adoption overall, and it’s clear: people are fine chatting with a bot, but handing it a credit card is another matter entirely.

Think about it from a regular person’s perspective. Asking an AI for a dinner recommendation carries basically zero risk. Letting it charge your card is a different kind of decision. If something goes sideways—wrong item, duplicate charge, or some subscription you never agreed to—who’s on the hook? Can it be undone? Is your financial information actually safe in the process?

Those aren’t unreasonable questions, and right now most people don’t have confident answers to them.

People still trust the brand behind the transaction.

Here’s where it gets interesting: Visa found that 61% of consumers said they’d trust Visa specifically to handle a payment initiated by AI. That number jumps to 68% for people between 18 and 34 and climbs even higher to 71% among people who already use AI frequently.

In other words, trust isn’t really about the AI at all. It’s about what’s standing behind it. People seem far more willing to accept AI-driven payments when they already recognize a name and trust a bank, a card network, or whatever is still part of the equation.

So maybe the real story isn’t “Will people trust AI with money?” but “Will people trust AI when it’s operating inside a system they already know?”

Trust, not tech, may decide how this plays out.

Visa Group President Oliver Jenkyn has compared this moment to the early days of e-commerce and mobile payments, a shift that could reshape how people shop and pay entirely. But he’s also been clear that none of it works without trust. That’s the real bottleneck.

AI agents could, in theory, do a lot: track prices, flag deals, manage recurring subscriptions, and even handle purchases automatically based on rules you set ahead of time. Convenient, sure. But convenience isn’t the whole pitch. People are going to want actual control visibility into what an agent is allowed to do, hard spending limits, and a say in when something needs their sign-off before it happens.

What payment companies will need to get right

A few things stand out as non-negotiable if agentic commerce is going to earn people’s trust:

Consumers need to know, clearly, when they’ve actually given an AI agent permission to spend on their behalf—no ambiguity. They’ll want spending caps they can set themselves. They’ll want to understand why an agent chose one product or transaction over another, not just get a receipt after the fact. Fraud protection and authentication will need to be rebuilt with AI-driven transactions specifically in mind, not bolted on as an afterthought. And when something goes wrong because something eventually will, there needs to be a real path to dispute it and get it fixed.

These aren’t nice-to-haves. They’re probably going to matter as much to consumers as speed and convenience do today.

The gap is a problem but also an opening.

Here’s the upside: that 72%-to-23% gap isn’t just a warning sign; it’s a market opportunity. Whoever figures out how to combine AI’s convenience with the kind of security people already trust could end up owning a huge chunk of this next wave.

Banks, card networks, digital wallets, and payment processors—they’re all positioned to become the trusted middle layer between AI agents and people’s actual money. The winning approach probably isn’t AI replacing the financial system as we know it. It’s AI working through the systems people already understand and rely on.

Where things go from here

Visa’s numbers paint a pretty clear picture: AI is everywhere, but AI holding the purse strings is still a hard sell. Closing that gap is going to be one of the defining challenges and opportunities in payments over the next few years.

Making AI smarter isn’t really the bottleneck anymore. Making it trustworthy enough to handle actual money is. For banks, fintechs, and payment providers, that’s the problem worth solving—and whoever solves it first will likely define what agentic commerce looks like for everyone else.

72% of US consumers have used an AI assistant—Visa Trust Index

👉 Follow us on Telegram: TheFinTechExperts

#PaymentsAgentic CommerceAIAI agentsArtificial IntelligenceConsumer Trustdigital paymentsfintechFintech Innovationgenerative AIpayment technologyVisa