Personal AI Agents Strain a Web Built for Humans

The agents do not fit existing categories. They are not human, and they work at machine speed. They are also not ordinary bots, because they act for individual people on tasks that can be legitimate and useful. Many websites have nevertheless blocked them because existing systems for telling bots from people cannot classify them.

TechCrunch reported that users had trouble deploying agents on sites including Pizza Hut and Yelp, which does not allow AI to book restaurant reservations. Amazon has blocked Muse’s agents and is fighting Perplexity in court to block its agents. A judge recently ruled for Perplexity, and Amazon has appealed, arguing that AI agents threaten its advertising business by making it harder to distinguish ads seen by humans from those served to bots.

Ethan Mollick, an associate professor at the Wharton School of the University of Pennsylvania, predicts the problem will soon affect the whole web. Once it becomes common knowledge that AI agents can be sent to request refunds or negotiate prices, Mollick wrote on LinkedIn, agents will overwhelm customer service departments to the point that they become unusable by real people. He also wrote that humans will eventually get very annoyed when their agents are not allowed to do this on their behalf.

Platforms have financial reason to admit agents. Most agentic web traffic goes to e-commerce sites, and Shopify is already welcoming agents by giving them their own checkout lanes that bypass pages designed for humans.

The question is what happens to the web for other users. MIT Technology Review describes a possible outcome: websites become neglected destinations for people, humans who browse and compare products face worse interfaces, verification systems remain imperfect and people are constantly asked to prove they are human, perhaps with an iris scan. Prices could vary by person’s agent, and users may not know when an agent’s recommendation is influenced by ads or platform commissions.

That outcome differs from the future promised by Google, Anthropic and OpenAI, in which agents give consumers more power and bargaining leverage. The article says that promise is not outlandish: agents can help people fight bogus medical bills or research cheaper car insurance. Whether they do so depends on rules still being written for how agents interact with websites and what they are allowed to do. Some rules could benefit everyone; neither consumers nor businesses want agents to wave credit card information around as they roam the web, the article says.

Editor’s Summary: Personal AI agents are now reaching users at scale, but many websites are blocking them because current systems cannot distinguish them from bots or people. Platforms such as Amazon and Shopify are taking opposing approaches, while experts warn that redesigning the web for agents could make it worse for humans. The rules governing agent-web interactions remain under discussion.