title: "What changed in Google and AI search: September 21, 2026"
date: 2026-09-21
version: 1.6.9
slug: seo-ai-search-news-september-21-2026
description: "No new SEOCode checks this week, but four things changed in the wider SEO and AI search world that are worth knowing about."
tags: [news, ai-search, llms-txt, open-graph, crux, webmcp]
What changed in Google and AI search: September 21, 2026
No new checks landed in SEOCode this week. But four things changed in the wider world of SEO and AI search that could affect how your site shows up, or gets found by AI tools.
What changed in Google and AI search
A detailed guide for helping AI agents use your website
A new protocol called WebMCP lets websites give AI agents a clear set of tools to work with, instead of making the agent guess how your site works by reading the raw page code.
Think of it this way. Right now, an AI agent visiting your site is like a person trying to use a vending machine with no labels. WebMCP adds the labels. You tell the agent: here is how to search our products, here is how to book an appointment, here is how to look up a document.
A full guide on how to set this up is now publicly available from DebugBear.
What to do: If your site lets people book things, look up products, or search documentation, it is worth reading the WebMCP guide. Setting up WebMCP endpoints could make your site much easier for AI agents to use correctly. This requires a developer to set up on the server side.
Payment pages can now get richer social previews
Open Graph is a system that controls how your pages look when someone shares a link on social media. You have probably seen it: a link shared on LinkedIn or Facebook shows a title, a description, and an image. That is Open Graph at work.
The Open Graph standard has added a new type called "payment". If your site has pages where people pay for something, you can now tag those pages with extra details: the amount, the status, an expiration date, and more. This makes those pages show up more clearly when shared on social platforms.
This is not a Google ranking signal. It will not help you rank higher in search results. But it can make your payment links look more professional and trustworthy when shared.
What to do: If your site has payment link pages, ask your developer to add og:type="payment" and the matching payment properties to those pages. If your site has no payment pages, you can ignore this.
Google is now tracking how ads affect your site's speed scores
CrUX is the Chrome User Experience Report. It is the real-world data Google collects from actual Chrome users visiting your site. Google uses some of this data to help decide how your site ranks.
Google has now added ad-related performance metrics to CrUX. This means you can see, in more detail, how the ads on your pages are affecting your site's speed and user experience scores. This is a reporting change, not a new ranking factor. Google is not saying ads will hurt your rankings. But the data is now there if you want to look.
What to do: Open PageSpeed Insights or Google Search Console and look at your CrUX data. If you run ads on your site, use Chrome DevTools to check how those ads affect your performance numbers. If your scores look bad, talk to your developer about ways to load ads without slowing the page down.
The llms.txt standard hit version 2, and it is now widely used
An llms.txt file is a simple text file you put on your website to help AI tools understand what your site is about and where to find your content. Think of it like a map for AI agents.
The standard just reached version 2. Here is how big it has gotten: thousands of sites now publish one, documentation platforms create them automatically, and Google's own Chrome Lighthouse tool now checks for them as part of a new "Agentic Browsing" category. Even AI labs publish their own llms.txt files.
Version 2 adds a few new features, including a way to link your regular web pages back to your llms.txt file so AI tools can find it more easily.
To be clear: Google Search does not use llms.txt to rank your pages. But outside of Google, AI agents and coding tools do use it to find and understand your content.
What to do: Consider adding an llms.txt file to your site. The format is simple: a heading with your site name, a short summary, and a list of links to your key pages or documents in markdown format. Ask your developer to also add a rel=describedby link to your HTML pages so AI tools can find the file. This will not change your Google rankings, but it may help AI tools find and use your content correctly.
Sources
- The /llms.txt Specification & Protocol
- DebugBear Technical SEO & Performance Lab
- Google AI Overviews & Generative Search Documentation
- ToTheWeb GEO & Generative Engine Research
- Web Vitals Metric Specifications (W3C / Web.dev)
- The Open Graph Protocol Specification
Catch SEO issues before they ship
SEOCode reviews every pull request against the latest Google SEO rules, automatically. No manual audits. No surprises after deploy.
It checks for missing metadata, broken structured data, hreflang errors, robots.txt issues, and more, right inside your existing GitHub workflow.
Want to see how your site is doing right now? Scan your website for SEO problems, free. No sign up.
Top comments (0)