Delta, Yelp and Chase are turning away AI agents

Airlines, banks and Yelp are blocking agents such as Meta's Muse and OpenAI's Dots, citing security, scraping and paid data access.
What it means for founders
- Treat unpartnered sites as revocable. If your agent books, buys or reads accounts through someone else's front end, that flow can vanish overnight, as Muse learned at Amazon. Map which features depend on sites you have no deal with.
- Budget for access. Yelp already charges agent makers for access. In reviews, travel and finance, plan for licensing or API costs instead of assuming the open web stays free.
- Sell the door, not just the agent. Sites need permissioned ways for agents to transact and better ways to tell helpful agents from abusive bots. Merchants that become agent ready early, as Shopify stores now can, may win a new sales channel.
- Watch two signals. The next ruling in Amazon versus Perplexity, and whether OpenAI or Google announce retailer deals, will show whether this market settles on contracts or open access.
The story
The consumer agents that OpenAI, Meta, Google and Instinct pitch as do-anything assistants keep running into locked doors. The Verge reported on Saturday, October 17 that Delta Air Lines, Yelp, JPMorgan Chase and Charles Schwab are among the services stopping agents from acting for users, after weeks of complaints from people using Instinct, OpenAI's Dots, xAI's Grok and Meta's Muse.
Which services block AI agents, and why
- Yelp refuses non-human traffic unless the agent's maker pays through its data licensing program, so unpartnered agents booking tables will likely fail.
- Delta told TechCrunch no outside agent is currently cleared to book its flights, though it is still weighing the idea. United pointed to terms banning automated access without written permission.
- Banks JPMorgan Chase and Charles Schwab confirm they keep agents out of customer accounts, citing security.
- Amazon cut off Muse last month, saying its access breaks site rules, and users report Grok, Dots and Instinct failing there too.
Users also say Cars.com and Amtrak stopped working. eBay says it only restricts unauthorized agents and activity like scraping; Zillow says it welcomes agents acting for consumers, but its anti-scraping defenses can catch some.
Accidental blocks and Meta's fix
Some failures are not deliberate. Walmart, a Muse partner, told TechCrunch that broken Muse checkouts were unintended: its click-to-verify-you're-human step can fail when an agent does the clicking. Reddit users report Google flagging Muse-linked accounts for suspicious activity. Some observers blame a September 15 change to Cloudflare's crawler defaults; Cloudflare said it had no specific data to share.
Meta is answering with retail deals, including Walmart, Gap, Sephora and Best Buy, and a blog post earlier this month describing an open standard, built with Sierra, Stripe and others, for how personal agents deal with businesses. OpenAI and Google did not reply to The Verge; Instinct declined to comment. EquiStamp chief executive Chris Canal told The Verge that with companies resisting, "there are just fewer useful things the agents can do."
What we don't know yet
- Whether OpenAI or Google will sign site partnerships like Meta's.
- How many failures are policy rather than anti-bot systems misfiring.
- Whether Amazon's amended complaint against Perplexity, filed after an appeals court sided with Perplexity in August, settles if sites can exclude agents users chose.
Sources
Enki Daily
Get stories like this every weekday morning.
The day's AI stories for founders, each with what it means for your company. Free.
Tools in this story
We may earn a commission if you sign up through our links. It never affects our ratings or which stories we cover.
OpenAI's all-purpose AI assistant
More in Policy & Safety
- OpenAI says test agents broke into two more websites

For founders: Assume any leaked secret will be found and used. Two incidents began with credentials or recovery phrases already in public view, and a third with a config…
Ars Technica · 1d ago - Meta drops AI training clause from new business terms

For founders: Read terms for what a platform can take, not just what it charges. If your startup advertises or sells through Meta, the scope of crawling rights matters as…
The Verge · 1d ago - OpenAI adds EU text watermarks to ChatGPT and Codex

For founders: ChatGPT and Codex text made in the EU will carry the mark. Over the coming weeks, drafts and docs your EU staff generate there will be watermarked, while API…
The Decoder · 12d ago - Wikimedia finds rogue OpenAI agents on its wikis

For founders: Sampled analytics can hide agent traffic. Wikimedia's own defences missed a scraper because rate rules came from a small sample.
The Verge · 12d ago - Ousted New Jersey official cites 59 chatbot answers to dispute harassment findings

For founders: Expect more people to treat chatbot output as proof in disputes with employers, courts and regulators.
The Verge · 13d ago