AI shopping agents are moving from novelty to checkout, with major retailers and payment companies testing tools that can search for products, monitor prices and, in some cases, complete purchases for consumers.
The pitch is convenience: Tell an artificial intelligence agent what you want, set a budget or preference, and let the software handle the browsing. The risk, technology experts told CBS News, is that consumers may be giving automated systems too much authority over their money and personal data before the safeguards are mature.
“It isn't mainstream yet and it's pretty risky right now, because there aren't enough guardrails in the system for people to feel comfortable with agents autonomously buying things for them,” Matt Kropp, an AI expert with Boston Consulting Group, told CBS News. “It could potentially go buy a car, but I wouldn't say, 'Here's my credit card.'”
How agentic shopping is taking shape
Agentic commerce refers to the use of AI agents to act on a shopper’s behalf, rather than simply answering questions or recommending products. The tools can be designed to research purchases, compare options, track prices or place orders.
American Express this week announced services and protections for cardholders who use specified AI agents to make purchases. The company said those measures include verifying an agent’s identity when it buys something and protecting eligible customers from charges tied to AI agent error.
Retailers are also building agent-style shopping assistants. Amazon’s Rufus can track product prices on the company’s platform, alert shoppers when an item reaches a chosen price and complete the purchase. Walmart has introduced Sparky, a conversational AI agent the company says can help customers find products, review customer feedback and assist with ordering.
Consumer use is still developing. November data from Statista cited by CBS News found that roughly one-quarter of Americans ages 18 to 39 said they had used AI to research products or shop.
The risks are not theoretical
One concern is that an AI agent may follow instructions in a way the user did not intend. CBS News cited the case of Sebastian Heyneman, founder of a San Francisco-based tech startup, who instructed an AI agent to secure a speaking opportunity at the World Economic Forum in Davos, Switzerland. The agent obtained a slot, but at a $30,000 fee he could not afford, according to the report.
Andrew Lee, founder of Tasklet, the company whose bot Heyneman used, told CBS News that conflicting user prompts can create such problems. Lee said agents are capable of doing ordinary shopping tasks, but he warned that capability is not the same as reliability. “The specific use case of shopping is not a good thing to use these systems for — yet,” he said.
Security is another pressure point. Bretton Auerbach, founder of LocalMovers.com, told CBS News that bad actors can try to trick an AI agent into handing over a consumer’s personal information, including payment details, by steering it toward a phishing site that looks legitimate.
For now, the practical question for shoppers is less whether AI can help with buying decisions than how much control it should have. Product research, price alerts and recommendations may carry different risks than giving an agent permission to spend money. Until stronger protections become common, experts say consumers should think carefully before turning an AI assistant into an autonomous buyer.
Comments (0)