What changed
Amazon, JPMorgan, Kohl's, Neiman Marcus and Schwab block personal AI agents; 37 of 51 large US retailers block them by accident; Walmart opens its site to agents
When
CNBC report November 1, 2026; Billions Network audit published October 6, 2026 (rules checked through September 2)
Who it affects
Any brand with its own site, plus Amazon and Walmart marketplace sellers whose discovery now runs through each platform's own agent
What to do
Decide your agent policy on purpose: read robots.txt and terms, pick a bucket, then measure agent referrals

Key takeaways

  • Amazon, JPMorgan, Kohl's, Neiman Marcus and Schwab block personal AI agents; Walmart welcomes them.
  • Of 51 large US retailers audited, 37 block shopping agents by accident and none lets a verified agent buy.
  • Decide your own agent policy on purpose, track AI referrals, and build listings Amazon's own agent can read.

How the agent wall went up

  1. Amazon adds six AI crawlers to robots.txt

  2. Amazon sues Perplexity over the Comet browser

  3. Ninth Circuit vacates Amazon's injunction

  4. Amazon cuts Meta's Muse off from buying

  5. Billions Network publishes the 51-retailer audit

  6. Walmart opens Agent Gateway to AI agents

  7. CNBC: JPMorgan and Amazon keep blocking agents

What happened

CNBC's Kevin Williams reported on November 1, 2026 that big banks and retailers are moving to block personal AI agents even as shoppers say they want to use them. JPMorgan CEO Jamie Dimon said the bank blocks external agents from acting for retail customers, citing security. Amazon has blocked Meta's Muse and other outside agents, and a spokesperson said Amazon asked Meta to remove Amazon from the Muse experience, adding that third-party shopping apps should operate openly and respect merchants' decisions about taking part. Charles Schwab, Neiman Marcus and Kohl's have moved the same way. Walmart is going the other direction and says agents will see the same prices as human shoppers. Chewy blocks an Amazon bot by name.

The Amazon position is not new. PPC Land traced it on October 8: Amazon added six AI crawlers to its robots.txt on August 21, 2025, sued Perplexity on November 4, 2025, won a preliminary injunction on March 9, 2026, then lost it on August 4, 2026 when the Ninth Circuit ruled that the user, not the AI company, was accessing Amazon's computers, while leaving Amazon free to police access through its terms. Meta released Muse on September 8, 2026 and Amazon cut it off on September 20, telling Muse users that an unauthorized agent violates Amazon's Conditions of Use. Andy Jassy has said Alexa for Shopping will be the main agent customers use on Amazon, and the company says more than 400 million customers used it in the last 12 months.

The audit numbers nobody else is quoting

Billions Network and its affiliate AmanTrue audited the published robots.txt files and terms of 51 of the largest US online retailers, drawn from the NRF Top 100 and the major marketplaces, in August 2026 and verified them through September 2. Only 11 of the 51 had taken a deliberate position on AI agents, split between restricting, limiting and welcoming them. Five, including Amazon, Kohl's and Neiman Marcus, explicitly restrict agents. The other 37, about three quarters, block shopping agents incidentally, because old rules that stop scrapers also stop a well-behaved agent at sign-in or checkout. Zero of 51 publish a route for a verified agent to buy, and only two are starting to let agents into checkout at all. Pottery Barn, West Elm and Williams-Sonoma publish llms.txt files to guide agents, then block checkout anyway.

Two caveats. The audit reads stated policy, not technical enforcement, so a retailer that blocks on paper may not block in practice, and the reverse. And Billions sells identity tools for AI agents, so its recommendation to build a verified-agent route is also its product pitch. The counts themselves come from public files anyone can check, and the dataset is published as a spreadsheet.

51 large US retailers, by agent policy

11took a deliberate position
5explicitly restrict agents
37block agents by accident
0let a verified agent buy
Source: Billions Network, October 2026

So what: accidental blocking is a sale you never see

A brand with its own site has one question to answer: when a shopper's agent lands on your product page, does it get to the cart? If your robots.txt still carries a blanket block on automated clients at the cart or sign-in path, the agent fails quietly and moves to a competitor or to Amazon. No dashboard reports that loss, which is why 37 retailers are blocking without having decided to. The demand side is real: Accenture research cited by CNBC found 65% of consumers are open to a GPT-style assistant acting for them, 82% want to approve every action and 79% want a stop button, which is exactly the human-review step Shopify now requires of agents that buy for shoppers, as covered in our Shopify Shop MCP checkout post.

On Amazon the decision has been made for you. Outside agents are cut off, so discovery runs through Alexa for Shopping, Rufus-style answers and sponsored placements, and a listing that Amazon's own agent cannot parse is the listing that loses. The Alexa for Shopping guide explains what the agent reads. On Walmart the opposite applies: agents are welcome, and your content has to be good enough for one to pick it. The AI agents for ecommerce guide covers the three cases side by side.

Worked example with illustrative numbers

These numbers are invented to show the arithmetic. A DTC brand gets 100,000 site sessions a month, converts 2.5% at a $60 average order, so $150,000 a month. Suppose agent-driven visits reach 4% of sessions over the next year, or 4,000 a month. If an old bot rule stops those sessions at the cart, they convert at 0%. At the site's normal 2.5% they would have produced 100 orders, or $6,000 a month and $72,000 a year. The fix is a few hours of developer time to read the robots file, allow the named agent user-agents through product and cart pages, and keep a human-review or spend-cap step at checkout. Even if agent traffic only reaches 1% of sessions, the fix pays for itself in the first month.

What sellers and brands should do

Each step names who does it, where, and the metric that shows it worked.

  • Web or dev lead: open your robots.txt and terms, search for GPTBot, ChatGPT-User, ClaudeBot, Claude-User, PerplexityBot, Google-Extended, GoogleAgent-Shopping, Meta-ExternalAgent and Amazonbot, and write down which of the audit's five buckets you are in. Metric: a dated, written agent policy.
  • Ecommerce manager: if you are on Shopify, read the robot and agent policy, enable Shop Pay agentic checkout where available and keep human review on. Metric: agent-sourced orders appear in reports with a zero or near-zero chargeback rate.
  • Analytics owner: add referrer segments for chatgpt.com, perplexity.ai, meta.ai, gemini.google.com and copilot.microsoft.com, with sessions, conversion rate and revenue by month. Metric: agent referral share and conversion rate, reviewed monthly.
  • Amazon account owner: stop planning around outside agents on Amazon and fill every structured attribute, A+ module and product Q&A so Alexa for Shopping can answer with your ASIN. Metric: Listing Quality Dashboard score and unit session percentage on your top 20 ASINs.
  • Walmart seller: complete item content for Walmart's agent channel the same way, since Walmart has said agents see the same prices as humans. Metric: Listing Quality Score in Seller Center and GMV by traffic source.
  • Owner: set a spend cap and a single-order limit for any agent checkout before opening it. Metric: fraud and chargeback rate on agent orders stays at or below site average.

Your agent policy in a week

  • Read robots.txt for the named agent user-agents
  • Pick one of the five audit buckets, on purpose
  • Allow product and cart pages, keep human review
  • Add AI referrer segments to analytics
  • Fill structured attributes for Alexa for Shopping
  • Set a spend cap before opening checkout

When this does not apply

If you sell only on Amazon and have no site of your own, the robots file is Amazon's problem, not yours, and your work is the listing content above. If your store is small and gets no AI referrals yet, do the audit of your own files, which costs nothing, and hold the development spend until the referrer segments show traffic. If you are in a regulated category where age or identity checks sit in checkout, keep the human step and treat agent checkout as a later project. And remember the audit measures policy, not enforcement: test your own store with an agent before you trust the file.

Where PPC Spider helps

We audit robots files, listing attributes and AI referral tracking as part of AI marketing services, and the brand mentions in AI search guide shows how to measure visibility inside the assistants. Follow the ecommerce marketplace news channel for the next agent policy change, or request a free Amazon audit to see how your listings read to Alexa for Shopping.

Sources

Get the next change before your competitors do

We send the marketplace changes that matter each week, with what to do next.

By subscribing you agree to our privacy policy. Unsubscribe any time.
Want this done for your account?

We run these playbooks on client accounts every day. Start with a free audit.

Get a free audit