Autonomous AI agents are being sold as the next big thing: assistants that book your dinner, clear your inbox and buy your shoes while you get on with your life. But as Emma Roth reports for The Verge, the open web is pushing back. In recent weeks, users of Meta's Muse, xAI's Grok, OpenAI's Dots and Instinct's agent have reported blocks stopping them from completing tasks on major websites.
The list of holdouts is growing. Yelp says it only allows agents that pay for its data licensing program, Delta Air Lines is blocking agents from booking tickets on users' behalf, and JPMorgan Chase and Charles Schwab have confirmed they block agent access to accounts over security concerns. Some users say Cars.com and Amtrak, which previously worked with their agents, are no longer accessible, suggesting more companies may be quietly closing the door.
Amazon vs. Perplexity: the legal battle behind the blocks
Amazon fired the first shot last year when it banned Perplexity's agentic shopping feature and sued the startup. Amazon said it had "repeatedly requested" that Perplexity stop letting its agents shop on its site, arguing third parties buying on behalf of customers "should operate openly and respect service provider decisions." Perplexity shot back that Amazon simply doesn't want agents roaming its platform because it means fewer human eyes on its ads, and compared its agent to an extension of the user.
The case has swung both ways: a judge ordered Perplexity to stop, an appeals court overturned that ruling in August, and Amazon has since filed an amended complaint. Meanwhile, Amazon told Meta that Muse's "continued access" violates its site policies, and there are reports that Grok, Instinct and Dots are also locked out of Amazon. xAI owner Elon Musk responded by floating the idea that he "should start a competitor to Amazon."
Why blocking AI agents hurts the whole pitch
Every block chips away at the promise that agents can do anything a human can do online. "Does this impact the perceived value of the Anthropics and the OpenAIs? I would say absolutely," Chris Canal, CEO of AI safety evaluator EquiStamp, told The Verge. "If there are companies that are actively working to make that not happen, there are just fewer useful things the agents can do."
