AI Shopping Agents Face Fraud, Liability, Security Roadblocks

The promise and the problem

AI shopping agents that autonomously complete purchases remain largely theoretical despite growing consumer interest, according to industry leaders speaking at Fortune Brainstorm Tech. While AI models help users discover products, fundamental barriers prevent agents from actually transacting on behalf of customers.

Matt Maher, founder and CEO of M7 Innovations, identified three core obstacles: security protocols, absent agentic commerce standards, and retailer policies designed to block third-party shopping agents. Even product discovery falls short—Melissa Bridgeford, cofounder and CEO of Wizard Commerce, noted that ChatGPT provides specific product recommendations only 9% of the time when users ask about items like ski gloves.

Bridgeford criticized OpenAI's early commerce efforts, saying the company fumbled by abandoning its Instant Checkout feature, which allowed purchases directly through the chat interface. That pivot caused retail partners including Walmart to exit the relationship.

Why it matters

The liability gap represents a significant business risk that could determine which companies lead—or lose—in agentic commerce. Without clear legal frameworks, merchants face potential losses from fraudulent purchases while technology providers risk customer trust and regulatory backlash. The first companies to solve identity verification and fraud prevention will likely capture the emerging market.

The liability vacuum

Courtney Robinson, head of policy and communications at open finance platform Akoya, called liability "one of the biggest unsolved challenges" in agentic commerce. When an AI agent makes an unintended purchase, no standards exist to determine who bears responsibility.

"Regulation always follows innovation," Robinson said. "Liability is wide open right now and being negotiated company to company, but there are no standards around where liability sits when an agent buys something that maybe the user didn't intend or ask for."

Maher warned that legal protections through terms and conditions won't shield merchants from what he termed "perceptual liability"—customers will still demand refunds and complain to retailers when agents make mistakes, especially loyal customers who expect accommodation.

Security threats multiply

Norman Menz, CEO of cybersecurity firm Flare, warned that AI agents will "magnify the problem exponentially" in an environment already plagued by ecommerce fraud. He identified two threat vectors: bad actors hijacking legitimate agents to make fraudulent purchases, and criminals creating agents using stolen identities and payment credentials.

"The attack surface keeps expanding," Menz said.

Adam Winnick, cofounder and CEO of Finality, argued that new open-source standards for agent monitoring and identity verification are necessary. He suggested blockchain technology could enable such systems, though other solutions might emerge.

The challenge is timing. Standards development typically requires years, but consumers want AI shopping agents now. "I think there is going to be a demand in the market to adopt and allow for the continued use of [AI shopping agents] before we have a solution to solve for fraud," Menz said.

Early adopters forge ahead

Ben Leventhal, founder and CEO of Blackbird Labs, said his blockchain-based restaurant rewards program is close to enabling AI agents to search restaurants and make reservations. In dining, payment fraud poses less risk since customers typically pay in person, but identity verification remains critical.

Leventhal predicted merchants will absorb fraud risk initially, as they currently do with "card not present" transactions in ecommerce. He remained optimistic about adoption: "Innovators and entrepreneurs are going to find killer use cases and they are going to be impossible to resist."

These details were first reported by Fortune.