Personal AI agents are beginning to shop, book and negotiate online, and websites are already adapting or blocking them. MIT Technology Review argues that the next redesign of the web could make life easier for agents while making the human version more irritating.
Watch Desk analysis
What happened
AI agents acting for users do not fit neatly into the web’s old distinction between people and bots. They can perform legitimate tasks, but at machine speed, and some websites have responded by blocking them. MIT Technology Review cites trouble deploying agents on sites including Pizza Hut and Yelp, and says Amazon has blocked Meta’s Muse agents.
The article also points to a different response: Shopify is giving agents their own checkout lanes, bypassing pages designed for people. It says Muse was downloaded five million times in its first month and agentic web traffic tripled from August to September. Those figures are reported by MIT Technology Review, which presents them as signs that agent use is growing.
Why it matters
If agents become common customers, websites may start optimising for automated visits rather than for people browsing and comparing things themselves. The article sketches possible consequences: more human checks, different prices shown to different agents, and uncertainty over whether an agent’s recommendation is influenced by advertising or commissions.
There is a real upside, too. An agent might help someone challenge a medical bill or find cheaper car insurance. Whether that works depends partly on the rules websites set for agents, including what they may access and do. Those rules are still being written, and businesses have their own reasons to welcome some agents and block others.
Our read
This is not just a question of whether your agent can get past a CAPTCHA. It is a contest over the web’s default customer: the person clicking through a shop, or the software doing it on their behalf. The trick will be making agents useful without turning the rest of the internet into an obstacle course for people.
What to watch
- Whether more websites block agents or build dedicated routes for them.
- How retailers disclose advertising, commissions and differences in agent-facing prices.
- What limits users can set on an agent’s access, spending and actions.
Discussion spark: Should websites be required to serve AI agents acting for users, or should businesses be free to block them even when that makes online tasks harder for people?
Sources and evidence
- Making the internet better for AI agents might make it worse for everyone else (19 October 2026, 18:22 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.