AI Shopping Agents Are Knocking—But Walmart’s Exit, Fraud Risks, and Liability Gaps Mean No One’s Ready

Posted on by

(SeaPRwire) –   By: Oliver Hawthorne

AI shopping agents are coming. But no one is ready to handle them. That was the consensus at a Brainstorm Tech panel on the topic. Users want agents to complete purchases for them. But security protocols, missing standards, and retailer blocks get in the way.

Melissa Bridgeford of Wizard Commerce says ChatGPT only gives specific product recommendations 9% of the time. OpenAI fumbled its Instant Checkout feature. Walmart pulled out of the partnership because of that pivot. Courtney Robinson of Akoya notes liability gaps—who’s to blame if an agent buys something unintended? Matt Maher of M7 Innovations calls this “perceptual liability.” A loyal Gap customer will still demand a refund for an unwanted blazer bought by their agent. Norman Menz of Flare says agents will magnify fraud. Bad actors could hijack agents or use stolen IDs to create fake ones. Adam Winnick of Finality wants open-source standards for agent identity, maybe using blockchain. Ben Leventhal of Blackbird Labs is close to letting agents book restaurant reservations, but identity verification remains a problem.

Merchants will likely bear fraud risks for now, just like they do with card-not-present transactions. Leventhal thinks killer use cases will make agents impossible to resist. But standards take years to build. The industry will have to adapt fast—before fraud and liability issues derail the whole idea.

Author bio: Oliver Hawthorne, Principal Correspondent at an international technology review, covers AI and commerce trends.