AI Business · · 5 min read
Why Do Shoppers Trust AI to Find Gifts but Not to Pay for Them?
We'll explore the paradox of AI trust in shopping, how businesses are attempting to bridge the gap for financial transactions, and what this means for the future of AI in commerce.

As the holiday shopping season ramps up, millions are turning to artificial intelligence (AI) agents for help. Walmart, Shopify, and Meta have all developed their own AI assistants, ready to suggest the perfect gift or track an order.
A recent Sinch survey of 2,501 holiday shoppers found that only one person didn't want to interact with AI at all. In the U.S., a significant 76% of shoppers feel confident AI can handle order tracking, and 64% trust it to help them learn about products.
This enthusiasm for AI's assistance might suggest a complete embrace of the technology in our daily transactions. But peek behind the curtain, and a different story emerges when it comes to the wallet.
While AI is increasingly influencing our shopping decisions, the moment money changes hands, trust evaporates. Only 55% of consumers trust AI for payment and billing requests, and a mere 28% of U.S. shoppers are comfortable with AI making a purchase on their behalf.
This isn't just about holiday gifts; it's a fundamental challenge for businesses integrating AI into their operations. How do you leverage AI's power without alienating customers who are wary of its financial autonomy?
That raises three questions.
Question 1
Why Do Shoppers Trust AI for Some Tasks but Draw the Line at Financial Transactions?
The data paints a clear picture of selective trust. While 76% of U.S. shoppers are confident in AI for order tracking and 64% for product learning, trust drops sharply for more sensitive tasks. Only 58% believe AI can handle account changes well, and just 55% trust it with payment and billing requests.
A Visa survey from 2026 highlighted this further: 54% of U.S. adults had used AI for at least one shopping task, but only 23% would trust generative AI to handle payment transactions on their behalf. This is a significant 31 percentage point drop between using AI for general tasks and for payments.
Why this gap? It boils down to perceived risk and control. Recommending a product is low risk; making a purchase is high. Kantar's September 2026 Commerce Barometer found that nearly half of U.S. shoppers are comfortable with AI automatically applying deals, likely because it saves money without direct financial risk.
However, only 28% are comfortable with AI making a purchase. The fear of errors, fraud, or loss of control over personal finances is a powerful deterrent. Mastercard's survey revealed that two-thirds (66%) of respondents would want to approve a transaction before an AI agent uses their loyalty points or card rewards. This suggests a strong desire for final human oversight on financial decisions.
And that brings us to the second question.
Question 2
How Are Businesses Trying to Build Trust for AI Agents Handling Payments?
Recognising this trust deficit, businesses are working to integrate AI agents into financial workflows with safeguards. The key is distinguishing between an AI agent's ability to execute a task and its authority to initiate it.
For context, Google's agentic AI guidelines explicitly recommend human approval for 'consequential actions,' which definitely includes financial transactions. It's about ensuring a human is in the loop, especially for irreversible actions.
Take the live pilot launched in October 2026 by Visa, Bill (a payments company), and Billtrust. These authenticated AI agents can complete buyer-approved business-invoice payments. Crucially, they are not deciding *which* invoices to pay; they are executing transactions on behalf of businesses *after* approval.
This is like a trusted assistant at a kirana store. The owner might trust the assistant to count out change or even suggest items to customers. But for a large order from a supplier, or authorising a new credit line, the owner's explicit approval is always needed. The assistant executes, but the owner authorises.
Vibhuti Sinha, chief product officer at identity security company Saviynt, emphasises that CIOs need to differentiate between an agent's identity, its system access, and its authority to perform specific actions. Logs show what happened, Sinha noted, but auditors need to know why it was allowed. This granular control and transparency are vital for building trust, especially in a business context.
And that brings us to the third question.
Question 3
What Does This Mean for the Future of AI Agents in Business Transactions?
Despite current consumer hesitation, the trajectory for AI agents in transactions is steep. Gartner expects AI agents to intermediate a massive 90% of purchases by 2028, pushing over $15 trillion of spending through these AI agent exchanges. This isn't just about influencing decisions; it's about being the actual conduit for transactions.
In Asia-Pacific, Matthew Driver, executive vice-president at Mastercard, noted that AI agents already influence around 15% of transactions, though purchases completed start-to-finish by agents are less than 1%. A report by Mastercard and Temasek forecasts this agent-influenced share to reach 30% to 40% by 2030.
This means that while human approval remains critical for now, the sheer volume of transactions passing through AI will necessitate robust systems. Patrick Parker, chief innovation officer at identity governance provider Omada, stressed that restrictions on AI agent actions must be enforced at the time of action, not just described in policy. This moves beyond 'human in the loop' to 'human defines the loop' with automated guardrails.
For businesses, this isn't just about consumer-facing applications. It extends to business-to-business payments, supply chain management, and internal procurement. The ability to deploy AI agents that can securely and transparently execute approved transactions will be a competitive advantage, but only if the underlying trust mechanisms are rock-solid.
The Bigger Picture
The journey from AI as a helpful assistant to a trusted financial executor is complex. It's not enough for AI to be smart; it must also be secure, transparent, and accountable.
Businesses that master this delicate balance, leveraging AI's efficiency while respecting the human need for control and oversight, will unlock immense value. Those that don't risk being left behind.
The future of commerce isn't just about AI making suggestions; it's about AI making transactions, but only when we truly trust it with our money.
Quick Answers
Why don't shoppers trust AI with payments?
Shoppers perceive higher risk and desire more control over their finances. While they trust AI for tasks like product discovery or order tracking, they require explicit human approval for sensitive financial actions like making purchases or using loyalty points.
How are businesses making AI agents trustworthy for payments?
Businesses are implementing 'authenticated AI agents' that execute transactions only after explicit human approval. This involves differentiating an agent's identity, system access, and specific authority to ensure human oversight for consequential financial actions.
What is the future role of AI agents in transactions?
Gartner forecasts that AI agents will intermediate 90% of purchases by 2028, driving over $15 trillion in spending. While human approval remains crucial, the focus will shift to robust systems that enforce AI agent restrictions in real-time and provide transparency for auditors.
Sources
- Quartz21 Nov · 00:12 IST
- TechTarget20 Nov · 00:21 IST
Facts are credited to the outlets above. The analysis, calculations and Signal Check are by the Finnect AI Insights desk.
