Visa CEO: Consumers Use AI to Shop But Won't Trust Agents to Pay

Consumers are embracing artificial intelligence for product research and comparison shopping, but they're drawing a hard line when it comes to letting AI agents complete purchases on their behalf, according to Visa CEO Ryan McInerney.

Speaking at the Goldman Sachs Communacopia + Technology Conference on Tuesday, McInerney outlined a clear pattern: shoppers use large language models and AI platforms to identify and compare products, then navigate to sellers' websites themselves to finalize transactions. The reason for this behavior comes down to a single word, he said: trust.

Why it matters

The gap between AI-assisted shopping and autonomous AI payments represents a critical inflection point for commerce infrastructure. Payment networks, banks, and technology platforms are racing to build the trust frameworks that could unlock agentic commerce—but consumer skepticism about handing financial control to algorithms may slow adoption more than technical limitations will.

The Trust Deficit in Agentic Payments

McInerney cited survey data showing three-quarters of consumers don't trust agentic platforms to make autonomous payments using their money and financial information. The trust problem runs in both directions: sellers need assurance that agents entering their sites are legitimate and properly authorized by consumers, while consumers must feel confident entrusting agents with payment credentials.

When consumers were asked whether they would trust an agent to make payments if Visa were involved, 61% said yes—a figure that rose above 70% among consumers who use LLMs at least weekly. The data suggests brand reputation and frequency of AI use both influence willingness to adopt autonomous payment capabilities.

Fraud Prevention Moves Upstream

Visa is responding to the trust challenge by shifting its fraud prevention approach earlier in the transaction lifecycle. Rather than focusing solely on identifying fraudulent transactions after they occur, the company is developing products that address identity risk before theft leads to fraud.

"Identity has become a critical area of vulnerability," McInerney said, discussing Visa's planned acquisition of BioCatch. The behavioral biometrics company can help clients protect identities on mobile devices before identity theft results in fraudulent transactions. Cybersecurity now ranks as a top-three issue for every financial institution client Visa works with globally, according to McInerney.

Expanding Beyond the Payment Credential

Tokenization is giving Visa new distribution channels for security and identity services beyond the payment credential itself. The company has scaled tokens globally, creating what McInerney described as a platform for distributing risk, identity, and transaction solutions.

Visa has captured only "very low single digits" of the addressable markets it sees across issuer services, acceptance, risk and identity, and advisory services. The company identified roughly $2 trillion in remaining addressable market for consumer payments, much of it still represented by cash and checks worldwide.

McInerney also highlighted opportunities in Visa Direct, which reaches 18 billion endpoints across accounts, cards, and wallets, and in higher-value cross-border B2B payments. Stablecoins could serve cross-border use cases in approximately 50 countries where consumers and businesses want to hold U.S. dollars but face barriers to doing so through traditional bank accounts.

The company is applying AI internally as well, with teams producing 80% more code commits, reducing product design and build time by 80%, and developing features 65% faster.

These details were first reported by PYMNTS.