Consumers Want AI to Shop with Them, Not for Them

Two-thirds of shoppers across the U.S., U.K. and Australia are open to agentic AI tools — but new Commerce/PayPal research shows they want AI confined to research and comparison, not checkout. For retailers, that means building trust into discovery, not automating the buy button.

Recent research from Commerce and PayPal suggests that agentic shopping (AI assistants that can research, compare and even purchase products on a consumer’s behalf) is gaining real traction, though full autonomy remains a hard sell.

What did the Commerce/PayPal study find?

The study, conducted by Logica Research and covering 3,000 online shoppers across the U.S., U.K. and Australia, found that roughly two-thirds of consumers in each market are interested in trying the technology. Enthusiasm was highest in the U.S. where 67% expressed interest, compared with 64% in the U.K. and 63% in Australia.

Despite that appetite, the research makes clear that consumers want AI in a supporting role rather than as an autonomous buyer. Shoppers said they value agentic tools most for finding lower prices and discounts and for cutting down research time, but they overwhelmingly want that help during product research and comparison — not at checkout.

Why Don’t Consumers Trust AI to Complete Purchases?

Purchase confirmation, fraud protection and strong payment security emerged as the top factors that would make people more comfortable handing tasks to an AI agent. Sharon Gee, Commerce’s senior vice president of AI for products, said the opportunity in agentic commerce is real, but that consumers “aren’t ready to trust agents as an autonomous buyer.”

Security concerns were consistent across all three countries. In the U.S., 82% of respondents said agentic tools need payment security equal to or better than existing methods, while in Australia that figure was 79%; both markets cited fears about unauthorized purchases and data breaches as barriers to adoption. Interestingly, the most common reason Americans gave for resisting AI shopping tools wasn’t distrust but enjoyment — 49% said they simply like the process of shopping.

Trust in who provides these experiences also varies by country, with U.S. consumers spreading confidence across tech firms like Google and Apple, payment companies like PayPal, marketplaces like Amazon and Etsy and card networks like Visa and Mastercard.

How Should Retailers Respond to Agentic Commerce Demand?

For retailers, the findings point to a narrower, more deliberate rollout of AI shopping tools than some may have anticipated.

Rather than racing to automate checkout, merchants have an incentive to build agentic features into the discovery and comparison stages of the journey — the part of the funnel where consumers are actually asking for help — while treating full purchase automation as a longer-term bet contingent on trust.

Retailers that can visibly demonstrate fraud protection, transaction transparency and purchase-confirmation steps may gain a competitive edge in adoption, while those that push AI-driven autonomous buying too aggressively risk friction with security-conscious shoppers. The research also suggests a shift in competitive dynamics: since consumers’ trust is distributed across tech platforms, payment providers and marketplaces rather than concentrated with any single type of company, retailers may need new partnerships across these categories to credibly offer agentic experiences rather than building trust unilaterally.

Related article: Why Virginia Ranked #1 for AI Adoption in the U.S.

Street Talk is the weekly newsletter for retail and fashion executives who want analysis, not noise. New issues every Thursday. Subscribe today.