Autonomous shopping and booking assistants are running into a wall on some of the web’s biggest sites. Users of Meta’s Muse, xAI’s Grok, OpenAI’s Dots, and Instinct report blocked tasks this month. The failures are not random. Companies are deciding which software gets through the door.
Yelp says it only allows agents that pay for its data licensing program. Delta Air Lines is blocking agents from booking tickets on behalf of travelers. JPMorgan Chase and Charles Schwab confirmed they stop agents from reaching customer accounts at all.
Some sites changed course without telling anyone. Users say Cars.com and Amtrak used to work with their assistants and no longer do. That quiet reversal is the part that stings. A site can rewrite the rules overnight and no customer gets a notice.
Amazon has fought this battle the longest. It banned Perplexity’s shopping agent last year and sued the startup. A judge ordered Perplexity to stop, then an appeals court overturned that ruling in August. Amazon filed an amended complaint after the reversal.
Perplexity argued that Amazon wants human eyeballs on its ads, not helpful assistants. Amazon countered that companies buying on behalf of customers should respect site policies. Meta’s Muse then hit the same wall, and Amazon said its continued access violates those policies. Elon Musk responded that he might start a competitor to Amazon.
The damage reaches past one retailer. Chris Canal, chief executive of the AI safety evaluator EquiStamp, told The Verge that blocked access undercuts the central promise of agents. “One of the claims is that they’re going to permeate every aspect of our lives,” he said. If companies actively prevent that, agents simply do fewer useful things.
Meta framed the issue as lost revenue. “Turning away a personal agent means turning away the customer behind it,” the company said this month. It would rather strike deals with businesses than fight them.
Even where agents are welcome, anti-bot systems can trip them up. Login screens, CAPTCHAs, and device checks treat automated browsers as hostile by default. Agents that look like browsers still fail the tests that browsers pass.
Banks have the clearest reason to say no. An assistant holding a password can move money, and liability stays murky when something breaks. Charles Schwab’s chief executive said the client cannot control the agent and the broker cannot control it either. Until someone owns that risk, account access stays closed.
The result is a web that looks open to people and closed to software. Agents can summarize pages and compare prices, but they stall at the checkout counter and the bank login. Every blocked site narrows what the assistants can honestly claim to do.
Read the full report from The Verge. For more on the people warning about agent risk, see our story on fired OpenAI safety staff.
The Netherlands, for internationals
One email a week: the week's news for internationals in the Netherlands, new practical guides and what changed in the cities. No spam, unsubscribe any time.
