Websites Block AI Agents from Performing User Tasks

Websites are increasingly blocking autonomous AI agents from performing tasks on behalf of users. While companies like OpenAI, Meta, Google, and Instinct are developing agents designed to book reservations, manage inboxes, and purchase products, many popular web services are preventing this access.

Reports indicate that users of Meta’s Muse, xAI’s Grok, OpenAI’s Dots, and Instinct have encountered blocks on top websites. Specific examples include Delta Air Lines, which prevents agents from booking tickets, and financial institutions like JPMorgan Chase and Charles Schwab, which have confirmed blocks on account access due to security concerns. Other sites, such as Cars.com and Amtrak, have also become inaccessible to agents that previously worked. Yelp has stated it only permits AI agents that participate in its paid data licensing program.

AI agents are designed to act as extensions of the user, scouring the internet to execute complex workflows. Instead of a user manually navigating a site, the agent interacts with the web interface to complete a goal, such as shopping for shoes or managing a travel itinerary.

To facilitate this, some companies are choosing partnership over blocking. Meta has partnered with retailers including Walmart, Gap, Sephora, and Best Buy to allow its Muse agent to shop on their platforms. Similarly, Shopify has opened its network of storefronts to AI agents. These partnerships create a sanctioned path for agents to operate without triggering security protocols.

There are three primary obstacles preventing the widespread adoption of AI agents: legal disputes, business models, and technical safeguards.

First, legal battles are emerging over the right to use agents. Amazon previously banned Perplexity’s agentic shopping feature and filed a lawsuit. Amazon argued that third-party companies shopping for customers should respect the provider's decision to participate. Perplexity countered that an AI agent is an extension of the user's right to hire an assistant. While an appeals court overturned a ruling that stopped Perplexity's agents in August, Amazon has since filed an amended complaint.

Second, there is a conflict regarding advertising revenue. Perplexity claims that Amazon blocks agents because it reduces the number of human users seeing advertisements on the platform. This suggests that the removal of human "eyeballs" from the browsing process threatens the ecommerce business model.

Third, agents often trigger existing anti-bot systems. Many websites employ safeguards to block programs used for hacking or scamming. Because AI agents behave like bots, they trigger these same security flags. For example, some Reddit users report that Google has blocked accounts using Muse due to "suspicious activity," and others have faced checkout issues on Walmart despite the existing Meta partnership.

The utility of AI agents is limited by the willingness of website owners to grant them access, turning a technical capability into a business and legal negotiation.