AI agents are increasingly designed to navigate websites, fill shopping carts and book travel on a user's behalf. Their usefulness, however, depends on whether each website allows automated access. In some cases, companies deliberately block these tools; in others, security systems mistake legitimate assistants for unwanted bots.
Meta's Muse has encountered human-verification checks while shopping at Walmart, one of the assistant's retail partners. Walmart said those checks were not intended to block the assistant. Amazon, by contrast, has deliberately blocked rival shopping agents from using its site, according to TechCrunch.
The concerns are particularly serious when financial transactions are involved. During JPMorgan Chase's October earnings call, CFO Jeremy Barnum said some agents rely on shared login credentials and screen scraping to transfer funds, describing those methods as insecure. The bank blocks them where possible, and Barnum warned that customers who provide their credentials could be responsible if an agent performs an unwanted action.
Amazon is also disputing access by Perplexity's Comet AI browser. Amazon argues that the agent violates its access rules and could put customer data at risk, while Perplexity says users have authorized it to act for them. A federal appeals court vacated a preliminary injunction against Perplexity in August 2026, but the broader dispute has not been resolved. Amazon also operates its own Buy for Me feature, which can purchase products from other brands' websites.
Commercial policies are another factor. eBay's user agreement prohibits unauthorized buy-for-me agents unless the company grants permission. Delta told TechCrunch that it has no integration for third-party agents to book flights through its website, although it did not confirm whether it intentionally blocks particular agents.
Not every failure indicates a deliberate ban. Walmart's verification prompts can interrupt automated shopping sessions, while Zillow has acknowledged that its anti-scraping defenses sometimes identify agents that it otherwise permits. From a user's perspective, both situations can look like an assistant that simply cannot complete a task.
These checks work similarly to the challenges that can affect people browsing through shared VPN networks. Google says such networks may trigger reCAPTCHA when their traffic resembles automated activity, and AI agents can face comparable checks while following a user's instructions. Using a VPN does not automatically bypass a website's restrictions.
Cloudflare has introduced controls that allow website owners to distinguish between search crawlers, AI training bots and agents acting for users. Since September 15, new domains using Cloudflare have defaulted to allowing search crawlers while blocking AI training bots and agents on pages displaying ads. Site owners can change those settings, so blocking training bots does not necessarily mean rejecting shopping assistants.
Other companies are opting for more limited integrations. Shopify's checkout tools allow compatible browser agents to update checkout details and submit orders after the buyer confirms, while returning control to the shopper for authentication steps that require their involvement.
Sources: Engadget.