Humans in the background: personal AI agents usher in a third category of web visitor

By James O'Donnell (The Algorithm, MIT Technology Review) · Mon, 19 Oct 2026.

Until recently, internet users fell into two categories: bots and people. Websites have long allowed bots to perform automated tasks, such as scraping page content, but there are rules about what they can access and how they behave. The web is designed mainly for people, even if we sometimes have to solve puzzles and captchas to prove we are not bots.

That is being upended by the arrival of personal AI agents, capable of doing things online on the user's behalf, such as booking flights or hotel rooms. They have been rolled out at scale with apps such as OpenAI's Dots and Meta's Muse. According to the article, Muse was downloaded 5 million times in its first month and agentic web traffic tripled between August and September.

Mixing these agents with human visitors, on a web designed to prioritize people, has gone badly: many sites block the agents that consumers send to negotiate prices or plan trips. If demand for agents keeps growing, the author argues, the internet will increasingly be redesigned for them rather than for us.

O'Donnell argues that agents are really a third category of visitor: they are not human and use the internet at machine speed, but they are not exactly bots either, because they act on behalf of specific people for tasks that can be perfectly legitimate and useful; they simply work faster.

Because agents do not fit into the systems for telling bots from people, some platforms, especially the most security-conscious, have opted to ban them altogether. TechCrunch reported problems deploying agents on sites ranging from Pizza Hut to Yelp, which does not allow AI to book restaurant tables. Amazon has blocked Muse's agents and is locked in a legal battle with the startup Perplexity to block its agents as well. A judge recently ruled in Perplexity's favor, but 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 (University of Pennsylvania), predicts that these blocks, currently a nuisance for early adopters, will end up affecting the entire web. Once it becomes common knowledge that agents can request refunds or negotiate prices, they will swamp customer service departments to the point of making them unusable for real people. Mollick wrote on LinkedIn: “Humans will end up very annoyed when their agents can't do this on their behalf.”

If demand for personal agents grows, platforms will find a way to let them in. According to the text, the vast majority of agentic web traffic goes to e-commerce sites, so there is money to be made by letting legions of agents shop on people's behalf. Shopify is already welcoming them with open arms, giving them their own checkout lanes that bypass pages designed for humans.

The key question is what will happen to the internet for the rest of us. We have accepted a web that, even while exploiting our attention, at least had humans as its intended users. But it could give way to one optimized for agents that will eventually outnumber us, making it even more unpleasant for the humans who remain.

The author sketches a possible scenario: websites are neglected as destinations for real people; those who want to browse and compare products themselves find worse interfaces; verification systems are imperfect and you would constantly have to prove you are human (perhaps even with an iris scan); a sweater would not have a single price for everyone, but would show different prices to each person's agents; and you would never know whether an agent's recommendation is influenced by advertising or platform commissions.

That is a far cry from the future promised by Google, Anthropic and OpenAI, in which agents would give consumers more power and bargaining leverage. Even so, the author considers that promise not far-fetched: agents can genuinely help, for example by fighting fraudulent medical bills or finding cheaper car insurance.

Whether they can be used for those purposes depends on the rules set for how agents interact with websites and what they are allowed to do. Those rules are still being written. Some will benefit everyone: neither consumers nor companies want agents exposing credit card details all over the web, and rules that strengthen agent security are positive. The most contested ones, such as whether companies can block agents or prevent them from negotiating better deals for consumers, will determine whether the web becomes a highway of agents working for us or against us.

Note: the email also includes the 'Deeper Learning' and 'Bits and Bytes' sections, which are not summarized here, and a sponsored ad from MIT Sloan Executive Education, which is omitted.