Skip to main content
Back to News
news/AI Applications

eBay Quietly Shuts Its Listings to Third-Party AI Agents

eBay now blocks outside AI agents from reading its listings, enforcing dormant bot terms and deepening the retail split over agentic shopping.

Stefan Trbojevic

Stefan Trbojevic

11 November 20262 min read
LinkedIn
Frosted glass storefront panels glowing teal against a dark frame

The takeaway

Agent access to commerce sites is becoming a negotiated, per-merchant permission problem. Builders should prefer sanctioned rails, log blocks as distinct failures, and identify their agents.

Why it matters for builders

Agent access is a per-merchant product decision. Use sanctioned agent rails (Shopify agentic storefronts, Walmart gateway, retailer APIs), distinguish blocked responses from empty result sets in your code, and identify your agent in the user-agent string.

eBay Quietly Shuts Its Listings to Third-Party AI Agents

eBay has begun blocking outside AI agents from reading its marketplace. The Verge reported on Tuesday that a reader's Meta Muse agent, which had been compiling a daily digest of watch listings with prices and shipping details, suddenly stopped pulling anything from the site. Muse said it had been blocked. The Verge reproduced the behaviour with other searches, and in some cases the agent routed users to third-party sites that mirror eBay listings instead.

eBay did not deny it. A spokesperson told The Verge that eBay has "restricted agent access in certain cases, including unauthorized scraping and training on our data," while declining to say whether new limits were imposed this month. Meta declined to comment.

The rules existed, the enforcement just arrived

The block is not a surprise legally. eBay rewrote its user agreement earlier this year to prohibit "buy-for-me agents, LLM-driven bots, or any end-to-end flow that attempts to place orders without human review," as Ars Technica reported when the terms took effect in February. What changed is the layer being enforced: agents tolerated as read-only research tools are now treated as unauthorized scrapers.

Retail is splitting into two camps

Amazon blocks external agents outright and requires bots to identify themselves. JPMorgan Chase, Charles Schwab, Kohl's and Neiman Marcus have taken similar positions, CNBC reported, even as Accenture research cited there found 71% of customers would welcome an agent inside their bank's app. Walmart is opening its catalogue to agents, Shopify syndicates a billion listings into agent storefronts, and Target lets agents through on some requests and blocks them on others through ordinary bot detection.

Two glass data blocks separated by a translucent barrier

The volumes explain the caution. Uber told The Verge it fulfilled four rides through third-party agents in the third quarter and roughly 500 Uber Eats orders during Halloween week. DoorDash reported "several hundred" agent orders against 993 million total orders last quarter. Retailers are being asked to carry fraud and inventory-hoarding risk for a channel that still moves a rounding error of volume.

What this means for builders

Agent access is now a per-merchant product decision, not a technical detail.

  • Build on sanctioned rails first. Shopify's agentic storefronts, Walmart's agent gateway and retailer APIs exist because raw site access does not scale. TechCrunch's survey of the problem found the same pattern across travel, banking and retail: policy, not capability, is the blocker.
  • Design for refusal. An agent that cannot tell "no results" from "blocked by the merchant" silently returns worse answers. Log block events separately and surface them.
  • Treat identity as a feature. Self-identifying user agents and published agent terms are becoming the price of admission, the way robots.txt became table stakes for crawlers.

The agentic shopping demo is easy. Keeping the merchant's door open is the engineering problem.

Share𝕏

The Automation Brief

Read 5 AI stories instead of 50.

The essential moves in AI agents, models, automation and infrastructure — filtered for builders and operators, with the part that actually matters.

No noise. Unsubscribe anytime.

Editorial notes

Reported by

Stefan Trbojevic

Edited by

n8n Lab Editorial

Published

11 November 2026

Updated

11 November 2026

AI disclosure: AI assisted with research and drafting. Factual claims are reviewed by an editor.

n8n Lab is an independent service provider. We are not affiliated with, endorsed by, or sponsored by n8n GmbH. “n8n” is a trademark of n8n GmbH and is used here only to describe the platform-specific implementation and automation services we provide.