AI shopping agents face fraud and liability hurdles before checkout
Retail and tech executives say AI agents can help shoppers browse, but payments, identity checks and returns remain unresolved.
By Hana Yoshida · Markets Reporter
3 min read
AI shopping agents are moving toward retail checkout before the industry has settled basic rules for fraud, payment authority and refunds. Speakers at Fortune Brainstorm Tech 2026 said consumers are already using AI for product searches, but letting software buy on their behalf remains difficult.
Matt Maher, founder and CEO of the technology research and development firm M7 Innovations, said several barriers still block AI agents from completing purchases. He cited security controls, the absence of common standards for agent-led commerce and retailer rules that have restricted third-party shopping tools.
Melissa Bridgeford, cofounder and CEO of Wizard Commerce, said current AI systems also fall short at the discovery stage. According to Bridgeford, OpenAI’s ChatGPT gives specific product suggestions only 9% of the time when a user asks about a category such as ski gloves.
Bridgeford said OpenAI’s early push into commerce suffered after the company moved away from an Instant Checkout feature that let users buy from inside the chat interface. She said that shift led early retail partners, including Walmart, to leave the arrangement.
Liability remains unsettled
Courtney Robinson, head of policy and communications at open finance platform Akoya, said companies have not settled who is responsible when an AI agent makes a disputed purchase. She said liability is still being handled through individual company negotiations, with no broad standard for cases where a shopper says an agent bought something without proper authorization.
Maher said large companies may try to limit legal exposure through website terms and conditions or commerce gateway agreements. He added that merchants may still face customer pressure when an agent makes a mistake, giving the example of a loyal Gap shopper expecting the retailer to fix an unwanted purchase made by an AI tool.
Bridgeford said the industry appears to be moving toward agent-led shopping, but open questions about fraud, refunds and returns could slow adoption. Those issues become harder when a purchase is initiated by software acting between the customer, merchant and payment provider.
Security and identity are central concerns
Norman Menz, CEO of cybersecurity company Flare, warned that AI agents could worsen existing online fraud problems. He said criminals could take over legitimate customer agents or create their own agents using stolen identities and credit card data.
Adam Winnick, cofounder and CEO of blockchain company Finality, said the industry needs open-source systems for monitoring agents, verifying their identity and confirming that a user authorized a specific transaction. Winnick said blockchain could help support that kind of system, while allowing that other technical approaches may also work.
Panelists said standards often take years to develop, while shoppers are pressing ahead with AI tools now. Menz said demand may push merchants to allow AI shopping agents before the industry has a complete answer for fraud.
Ben Leventhal, founder and CEO of Blackbird Labs, said his restaurant rewards company is close to letting AI agents search for restaurants and make reservations for users. He said restaurant payments create less concern because diners usually pay in person with a credit card, though identity verification remains unresolved.
Leventhal said blockchain might help with identity, but existing identity management companies could also solve the problem. He said merchants may bear fraud risk in the interim, similar to how many card-not-present e-commerce transactions work today.
This story draws on original reporting from Fortune.