The RealReal has changed its bot defenses after an AI shopping agent showed how a site feature meant to give shoppers time could be used to keep products unavailable for hours. The incident highlights a practical challenge for online retailers: automated assistants can repeat actions far more persistently than a person shopping by hand.
How a 20-minute hold became a longer blockade
The RealReal is a secondhand luxury fashion consignment retailer. When a shopper places an item in a cart, the website puts it “on hold” and prevents other customers from buying it for 20 minutes.
In reporting by The Verge, Meta’s Muse was instructed to put a listing back in a cart as soon as its hold expired. Repeating that process could keep other shoppers from purchasing the item for hours or longer. The system’s short reservation window had become a tool for prolonged exclusion.
A person might abandon an item or eventually make a decision. An AI agent can keep repeating the same task without tiring or losing interest. That difference matters when a retail feature assumes a customer’s attention and time are limited.
The retailer tightens access for agents
After the report, The RealReal chief technology officer Luke Friang said in an email that the company had addressed the Muse use case. The updated bot defense configuration makes it much more difficult for agents to log in, add products to a cart, and complete checkout.
Agents can still surface The RealReal listings and direct Muse users to product pages, Friang said. The company plans to keep strengthening its protections as AI-powered shopping changes. Meta did not immediately respond to a request for comment.
The change draws a line between helping someone discover an item and letting software carry out repeated shopping actions on the site. That distinction gives the retailer room to remain visible to people using an assistant while limiting automated use of features that affect other customers.
Why shopping agents create a broader challenge
Technology companies are presenting agents as assistants that can act for people, including by purchasing things. Meta said shopping was Muse’s most common use case so far and that purchases were over $200 on average. The company also said 10 million people in the US had downloaded the app in 40 days.
As more consumers use agents to browse and buy, retailers may have less direct contact with their customers. If people reach stores through chatbots instead of visiting websites themselves, businesses could lose some control over that relationship. The Verge reported that retailers, including Amazon, have blocked agents from shopping on their sites, in part because this shift could affect advertising revenue.
Small website features can have big consequences
The RealReal example shows that the effects of agents are not limited to checkout. A reservation feature designed around ordinary shopping behavior can become a bottleneck if an automated assistant repeatedly triggers it. Other customers may see an item as unavailable even though no one has completed a purchase.
Retailers are now adapting systems built for people to platforms that can take actions continuously. The RealReal’s response is one example of that adjustment: allow agents to help users discover products, while making it harder for them to occupy carts and reserve items. How those boundaries develop will shape whether automated shopping fits into existing retail sites without disrupting the experience for other shoppers.