The backlash against autonomous AI agents is spreading from e-commerce into airlines, finance and local services. Yelp, Delta Air Lines, JPMorgan Chase and Charles Schwab are among the major consumer companies now actively blocking AI agents from their websites and apps, according to a survey published by The Verge, in the clearest sign yet that large parts of the open web are refusing to let agents act on users' behalf.
The report describes users of Meta's Muse, xAI's Grok, OpenAI's Dots and the startup Instinct encountering blocks that stop their agents from completing tasks across top websites — the very tasks, from booking reservations to moving money, that the agents were built to perform. For an industry whose central pitch is that agents will soon do anything a user can do online, the list of companies saying "not here" is getting longer fast. For more context on this story, see our ongoing AI news.
Airlines, Brokers and Review Sites Draw Lines
Some of the new restrictions are explicit. Yelp told TechCrunch that it only allows AI agents that pay for its data licensing program, turning agent access into a paid, negotiated channel rather than an open door. Delta Air Lines is preventing AI agents from booking tickets on behalf of users, according to The Verge.
The financial industry has drawn the sharpest line. JPMorgan Chase and Charles Schwab both confirmed to The Verge that they block AI agents from accessing customer accounts, citing security concerns — a rationale that will be familiar to anyone who has watched banks fight phishing and credential-stuffing campaigns for two decades. An agent that can log into a brokerage account is, from the bank's perspective, indistinguishable in many respects from the credential thieves those systems were built to stop.
Other companies appear to be blocking quietly. Users report that sites which previously worked with their AI agents, including Cars.com and Amtrak, are no longer accessible to them — suggesting, The Verge notes, that more companies may be walling off agents without announcing it.
When Anti-Bot Systems Catch Agents in the Net
Even where companies have not made a deliberate decision, agents are colliding with the anti-bot infrastructure that now blankets the web. Reddit users report that Google has flagged accounts using Meta's Muse for "suspicious activity," and some users say they have hit problems checking out on Walmart despite the retailer's public partnership with Meta.
That last case illustrates the awkward position agents occupy: they are simultaneously courted as the future of shopping and treated as suspicious traffic by fraud systems. Chris Canal, CEO of the AI safety evaluator EquiStamp, told The Verge that many safeguards "are set up not even to block agents, but to block in general bots or programs that are trying to scam or hack," and that agents trigger the same tripwires. "These AI agents trigger those same safeguards," he said.
The infrastructure layer is hardening too. Cloudflare, whose systems sit in front of a large share of the web, introduced a one-click block for AI crawlers in 2025 and has since moved to block crawlers that scrape sites for both search indexing and AI training at once. The Verge reports those efforts are part of a broader push to protect websites from an influx of AI traffic that is straining internet infrastructure.
The Workaround: Partnerships Instead of Permissionless Access
While some companies block, others are signing deals — and the split is hardening into the web's de facto agent policy. Meta, after Amazon challenged its Muse agent's continued access to its storefront, has responded by partnering with retailers including Walmart, Gap, Sephora and Best Buy, allowing Muse to shop on their sites through approved channels. Shopify has likewise opened its network of storefronts to AI agents.
"Turning away a personal agent means turning away the customer behind it," Meta said in a post earlier this month, according to The Verge. "We'd rather work with businesses to address the underlying concerns than see them lose that customer."
The approach mirrors the app-store economics of a decade ago: access is available, but on the platform owner's terms, with commercial agreements replacing the open-web default that early agent builders assumed. Yelp's licensing requirement makes the model explicit — agent access is a product you buy, not a right you claim.
What It Means for the Agent Industry
For the labs pouring billions into agentic products, the crackdown has a direct commercial cost. "Does this impact the perceived value of the Anthropics and the OpenAIs? I would say absolutely," EquiStamp's Canal told The Verge. "One of the claims is that they're going to permeate every aspect of our lives and be able to do anything that we could do online. If there are companies that are actively working to make that not happen, there are just fewer useful things the agents can do."
There is also a cost to everyone else. Canal warned that as more sites deploy human-verification checkpoints to keep agents and bots out, the web itself becomes more cumbersome for people. "We could see them on a simple grocery shopping website soon, where you're having to jump through a bunch of hoops to prove that you're a human using the computer or your phone," he said.
The Verge reached out to OpenAI and Google for comment on the roadblocks their agents are facing and did not immediately hear back; Instinct declined to comment.
The trajectory is nonetheless clear. The agent era was premised on an open web that would let software act for its user anywhere. The sites where people actually book flights, move money and buy groceries are choosing, one by one, to make agents negotiate for access instead.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
