AI Payments Uneasy Public
Visa plugged ChatGPT into your wallet. Most Americans aren't ready.
How consumers feel about Visa's ChatGPT payment integration
On this page
Share It On
Executive summary
Visa just wired its payment network directly into ChatGPT — and most Americans aren't ready for it. When nearly 200 consumers were surveyed about the June 2026 announcement, nearly six in ten said their first reaction was concern about security risks, while only one in five felt excited about the convenience. That gap between corporate ambition and consumer readiness is the central tension shaping the future of AI-driven commerce.
The numbers tell a consistent story across every question asked. Security dominates: 68% of respondents named fraud protection as their single biggest priority when using AI to make purchases — dwarfing deal-seeking (18.5%) and speed (7.2%). Open-ended responses show consumers aren't rejecting AI commerce outright; they're insisting that a human remain in the approval loop for every transaction. Personality research adds another layer — people who score high on emotional volatility are significantly less likely to embrace this integration, suggesting a segment of the population faces anxiety that goes beyond rational risk calculation. The opportunity is real and enormous, with McKinsey projecting up to $1 trillion in US agentic commerce by 2030. But the trust infrastructure has to be built first.
Context
On June 10, 2026, Visa announced it had embedded its payment network into ChatGPT, allowing the AI to browse, select, and complete purchases on behalf of users who link a Visa card. The integration uses tokenized credentials, real-time fraud monitoring, and user-defined spending limits — protections Visa's Chief Product Officer Jack Forestell described as the foundation for a shift bigger than the birth of e-commerce itself. Mastercard confirmed its own first agentic transaction took place in Q3 2025, signaling that the two dominant payment networks are racing toward the same destination.
This survey captured immediate consumer sentiment: 196 adults responded to four questions — two multiple-choice and two open-ended — in the days following the announcement. The sample is not nationally representative, but the distribution of responses closely mirrors larger industry benchmarks from Worldpay (surveying US adults in November 2025), Aevi (3,000 US and UK adults, April 2026), and Checkout.com (published the day before the Visa announcement in June 2026). The convergence across independent data sets strengthens confidence that the patterns here reflect genuine, durable consumer attitudes rather than noise.
The broader market frame matters: McKinsey estimates agentic commerce could orchestrate up to $1 trillion in US B2C retail transactions by 2030, with global projections reaching $3–5 trillion. Worldpay pegs US AI-agent-driven online spending at $261 billion by the same year. Against those projections, the consumer sentiment captured here — dominated by security anxiety and a strong preference for human oversight — represents the friction that will determine whether those numbers materialize on schedule or get pushed out by trust deficits. The question isn't whether agentic commerce arrives. It's how fast, and at what cost to consumer confidence.
Findings
Security anxiety is the dominant consumer response — by a wide margin
Almost six in ten respondents (57.9%) said they were "concerned about security risks" when they heard Visa had embedded its payment network into ChatGPT. That's nearly three times the share who said they were excited (20.0%), and it makes security worry the single largest reaction — not a plurality, but a near-majority.
The priority question confirms this isn't incidental. When asked what matters most when using AI to make purchases, 68.2% named "Security and fraud protection" — a figure that dwarfs "Getting the best deals" (18.5%) and reduces "Speed and convenience" to a distant third at just 7.2%. In a category that the payments industry has positioned primarily as a convenience story, consumers are telling a different story: safety first, everything else later.
These numbers don't stand alone. Worldpay found 48% of US adults call fraud protection essential to AI shopping adoption. Aevi found 59% fear losing control over how or when payments are made. This study's respondents are, if anything, more security-anxious than those broader benchmarks — a signal for any product team planning a consumer-facing rollout.
Takeaway: Biggest priority when using AI for purchases
Takeaway: Biggest priority when using AI for purchases
Desired Level of Control
Some respondents want checks and balances, while others prefer the AI to act without human intervention.
Hover over dots to see real answers.
Most respondents demand human approval before any AI purchase, with only a rare few willing to grant full autonomy.
Highlighted answers
- Require mandatory human approval for every purchase
“I would just want to ok the transaction first”
Distills the survey's dominant finding into its simplest form: a human must stay in the approval loop.
- Require mandatory human approval for every purchase
“It's pretty darn strange. I want to be able to select and approve my own purchases. I don't even trust Amazon subscriptions or Alexa to make purchases for me. I need to have control incase I change brand preferences”
Frames the control demand in everyday terms, showing distrust extends beyond new AI tools to established platforms.
- Require mandatory human approval for every purchase
“What if the AI just starts randomly buying things with my Visa card that I wasn't going to buy and may not be able to afford, what if the AI gets hacked while shopping and my Visa card information gets stolen. This feels like a gigantic security risk.”
Directly names the Visa-card scenario from the announcement, linking unauthorized spending and hacking fears to the article's central tension.
- Require mandatory human approval for every purchase
“The big concern is liability in that mistakes can be made, data privacy can come into play, as can it being easier for false identity. Transparency can also be a problem in that the AI may choose a specific price, product, or seller but does not make the reasoning behind the choices available to th”
Raises the trust-infrastructure gap the article identifies — liability, transparency, and privacy — as prerequisites before autonomy is acceptable.
- Allow AI to purchase autonomously for convenience
“None, if I am going to link my card then its like giving permission for AI to do this”
Represents the rare convenience-first minority who see card linkage itself as sufficient consent, anchoring the high pole of the spectrum.
Conclusion
The Visa-ChatGPT integration is the most visible test case yet for a technology that could reshape how trillions of dollars move through the global economy. But the consumer data says the industry is running ahead of the people it needs to bring along. Nearly six in ten Americans greet this development with security anxiety. Only one in five is excited. And across multiple independent surveys, the message is the same: slow down, show us the controls, and let us stay in charge.
Visa's technical architecture — tokenized credentials, spending limits, merchant category controls, real-time fraud monitoring, and its new Agentic Directory and Agent Score tools — maps directly onto what consumers say they need. The UK Competition and Markets Authority's March 2026 guidance, making businesses legally responsible for illegal actions by their AI agents, adds a regulatory floor beneath the consumer trust floor.
Watch for two signals in the months ahead: whether Visa and OpenAI surface those security features prominently in consumer-facing communications (they've built the infrastructure; now they need to market it), and whether adoption numbers among 18–34-year-olds — where Worldpay found 59% willing to let a smart bot browse on their behalf — diverge meaningfully from the broader population. The gap between a $1 trillion market projection and today's wary consumer is real. It will close through demonstrated trust, not through convenience promises.
Takeaway: What would be your biggest priority when using AI for purchases?
Security and fraud protection
Getting the best deals
Speed and convenience
Other
Takeaway: What would be your biggest priority when using AI for purchases?
See echo in five minutes.
Bring a question. Get a real answer from real people, on the AI they already use.