AI
AI Agents Face Growing Website Blocks as Retailers Grapple With Bot Access
As consumer AI agents like Meta's Muse expand their ability to shop and book services, retailers are blocking them at scale, forcing the industry to develop new standards for distinguishing trusted assistants from malicious bots.
TechCrunch — AI
A new wave of personal AI assistants—including Meta's Muse, ChatGPT's Dots, and others—promises to handle everyday tasks like flight bookings, dinner reservations, and grocery orders without requiring users to manage technical setup. Yet as these tools proliferate, they're running headlong into a contentious problem: websites that refuse to let them in.
The Blocking Problem#
Earlier this fall, Amazon made headlines by blocking Meta's Muse from accessing its shopping platform entirely. But Amazon's move is not an outlier. Across social media, users report encountering barriers when their agents attempt to make purchases or reservations at retailers ranging from Walmart and eBay to airlines like Delta and United. Some retailers appear to be intentionally shutting out agent traffic, while others unintentionally block it through standard anti-bot security measures designed to catch spam and malicious scripts.
The confusion works both ways. Consumers often cannot tell whether a block is deliberate, leaving them frustrated with both the retailer and the agent provider. Meanwhile, websites struggle to distinguish between trusted personal assistants acting on users' behalf and genuine threats.
Why Websites Are Blocking#
The barriers stem from overlapping concerns. Legacy security—CAPTCHA checks and automated bot detection—can interrupt agent transactions, accidentally locking them out. Walmart, which has actually partnered with Meta on Muse, found that its human-verification prompts were failing when agents tried to complete purchases. Airlines cite concerns over unauthorized access; United's terms of service explicitly prohibit automated tools from using its platform, while Delta says it would need to implement strict security controls before opening its systems to third-party agents.
Other platforms take a stricter stance. Yelp requires that agents pay for access through its data licensing program if they wish to search reviews or make reservations. eBay restricts unauthorized scraping and model training but hasn't ruled out legitimate agent transactions altogether. Major infrastructure providers also play a role: Cloudflare, which operates security services for many websites, recently adjusted its defaults to block AI training, inadvertently creating friction for consumer agents.
