· AI agents 2026 tracker · Updated weekly

The Agent-Access Index: Which Sites Allow AI Agents in 2026?

The Agent-Access Index is Ingenious AI's live tracker of which websites allow AI agents in 2026 — and which ones block them. October 2026 turned agent access into a boardroom issue: Yelp, Delta, JPMorgan Chase, and Charles Schwab moved within days of each other to block Meta's Muse, xAI's Grok, and OpenAI's agents, and Citibank and Bank of America joined the block list on October 18. Meanwhile Walmart, Gap, Sephora, and Best Buy signed deals to let Muse make purchases for shoppers.

This page tracks 20 sites across banking, travel, retail, and technology. Each row shows the site's agent policy, the agents affected, the source of the report, and the date it was last checked. Agent policies are changing fast — this index is updated weekly.

Bottom line: as of October 18, 2026, the web is splitting in two. Banks and travel companies are walling AI agents out over liability fears, while big retailers are cutting deals to let agents shop. Use the table below to see exactly where every major site stands.

What is the Agent-Access Index?

The Agent-Access Index is a site-by-site policy tracker for the AI agents 2026 era. A status of Allows means the site welcomes agents (or signed a deal with one). Blocks means the site actively refuses agent traffic. Paid tier means agent access exists but costs money — Yelp's model. Unknown means the policy has not been verified yet. The dataset behind the table is a plain JavaScript array, so each weekly update is a one-line edit.

Agent policies by site: the live table

SiteStatusAgents affectedSourceLast checked

Methodology: statuses compiled from technology-outlet reporting published October 17-18, 2026, plus platform announcements. These are reported statuses, not independently verified. Last full check: October 18, 2026.

Submit a policy change

Spotted a site that changed its agent policy? Send it in — include the site, the new status, and a link to the source.

Submissions are reviewed manually before they appear in the index. There is no automatic publishing and no fake confirmation emails — a human reads every submission.

What does the agent-blocking wave mean for AI agents 2026?

It means the open-web assumption behind most AI agents 2026 roadmaps is breaking. Agents were built on the idea that the whole web is browsable; October 2026 proved that banks, airlines, and review platforms will close the door when liability is unclear. JPMorgan CEO Jamie Dimon warned on October 13 about "liability transfer" — who pays when an agent books, buys, or moves money wrongly. A Charles Schwab executive put the control problem bluntly: "the client cannot control the agent and the broker cannot control it either."

For site owners, the strategic question is no longer whether agents are coming, but on what terms: block them, charge them, or sign them as a sales channel. Every bank on the block list — JPMorgan Chase, Charles Schwab, Citibank, and Bank of America — announced after Dimon's October 13 warning: 100% of the banking wave. Retailers chose the third option. Banks chose the first. Most sites have not chosen at all — which is itself a policy, and usually the riskiest one.

How to use the Agent-Access Index each week

  1. Check your category. Filter the table by your industry — banking, travel, and retail are moving fastest — and note which competitors allow or block agents.
  2. Read the source column. Every status traces to outlet reporting or a platform announcement. Treat unverified rows as leads, not facts.
  3. Decide your posture. Block, paid tier, or welcome. A deliberate policy beats the default of having none when an agent incident lands.
  4. Submit what you see. Agent policies change weekly. If a site flips, submit it and the index improves for everyone.

The Ingenious AI suite — 11 security platforms including agent monitoring, disclosure compliance, and attack-surface watch — opens early access to teams dealing with the agent era. Get early access on WhatsApp.

Checker: how deliberate is your agent policy?

Six scored questions. Deterministic self-assessment — scored against fixed best practices, nothing is sent anywhere.

1. Has your company decided its agent posture: block, paid tier, or welcome?

2. Do you know which AI agents (Muse, Grok, OpenAI agents) currently access your site?

3. Does your terms of service mention AI agents specifically?

4. If an agent transacts wrongly on your site, do you know who is liable?

5. Could you switch posture (for example block to paid tier) within a week?

6. Do you re-check agent policies at least monthly?

Frequently asked questions

Which major sites block AI agents as of October 2026?

As reported by tech outlets in October 2026: Yelp (paid tier for agents), Delta, JPMorgan Chase, Charles Schwab, Citibank, Bank of America, and Amazon (which blocked Meta's Muse in September 2026) all restrict AI agent access. These are reported statuses from outlet coverage, not independently verified.

Which sites allow AI agents to shop?

Walmart, Gap, Sephora, and Best Buy signed deals with Meta letting the Muse agent make purchases for shoppers, and Shopify opened its network to third-party shopping agents. Meta, xAI, and OpenAI welcome agents on their own platforms by definition.

What is the difference between 'blocks' and 'paid tier'?

Blocks means the site refuses agent traffic outright. Paid tier — Yelp's model — means agent access exists but must be purchased. A paid tier is a business model; a block is a closed door.

How often is the Agent-Access Index updated?

The full table is re-checked weekly. Breaking policy changes — like the October 17-18 banking wave — are added as they are reported. Every row carries its last-checked date.

Why are banks blocking AI agents?

Liability. JPMorgan CEO Jamie Dimon warned on October 13, 2026 about 'liability transfer' — nobody has settled who pays when an agent acts wrongly with money. Until that is answered, banks are choosing to block first and negotiate later.

Related checks