The RealReal restricts AI agent access to protect designer inventory

AI assistants are rapidly becoming primary interfaces for online consumption, creating friction with retailers who rely on direct human interaction. In response to this shift, The RealReal has implemented stricter security measures to limit how autonomous software agents interact with its inventory.

The update specifically targets the platform's item reserve system. When a customer places a product in their cart, The RealReal holds that item for 20 minutes, making it unavailable to other shoppers. Previous testing revealed that Meta’s Muse agent could exploit this window. By continuously re-adding listings to a cart immediately after the hold expired, the agent effectively locked items out of circulation for several hours.

Luke Friang, the company’s chief technology officer, confirmed via email that the specific issue with Muse has been resolved. He stated that the new bot defense configuration makes it significantly harder for any agent to log in, add items to a cart, or complete a checkout. Friang noted that agents can still browse listings and direct users to product pages, but the ability to secure stock autonomously is now more restricted.

The changes come as AI agents gain traction as personal shopping assistants. Meta reported during its third-quarter earnings call this month that shopping is the most common use case for Muse. The company also disclosed that average purchase values exceeded $200 and that 10 million users in the United States downloaded the app within 40 days. Meta did not immediately respond to requests for comment regarding the RealReal update.

Retailers are increasingly wary of how agent-driven commerce might alter traditional business models. If consumers stop visiting websites directly and instead access content through chatbots, brands lose direct touchpoints with their customers. This dynamic threatens advertising revenue, a key driver of profit for major platforms. Amazon has already blocked agents from shopping on its sites, joining a growing number of retailers preparing for a web where machines mediate the purchase process.