The Short Version
Google Search
Not needed. Not read.
The official Search documentation says no new AI text files are needed to appear in AI features. Google staff have said there are no plans to support llms.txt.
Google Chrome
Audited. Optional for now.
Lighthouse checks for llms.txt inside its new Agentic Browsing category. A missing file is Not Applicable, not a failure. A broken one is flagged.
Our answer: not for rankings, yes for agents, and only if the file is small. The reasoning and the numbers are further down.
What Google Search Says
Three statements, from the official documentation and from the two people at Google who speak most about how Search works. They all point the same way.
Google Search Central documentation
"You don't need to create new machine readable files, AI text files, or markup to appear in these features. There's also no special schema.org structured data that you need to add."
From AI features and your website, the page that explains how to appear in AI Overviews and AI Mode. "AI text files" is the closest Google's documentation comes to naming llms.txt.
Gary Illyes, Google Search Central Live, July 2025
"Gary Illyes clearly stated that Google doesn't support LLMs.txt and isn't planning to."
Reported by attendee Kenichi Suzuki from the Search Central Live Deep Dive event in Asia Pacific, and covered by Search Engine Land. Google has not contradicted it since.
John Mueller, April 2025
llms.txt is "comparable to the keywords meta tag", and "none of the AI services have said they're using LLMs.TXT, and you can tell when you look at your server logs that they don't even check for it."
Posted on Reddit and covered by Search Engine Journal. The keywords meta tag is a pointed comparison. Google stopped using it in 2009 because site owners could write anything in it.
That is the whole Search position. It is not hedged, and it has not softened. If your goal is to rank in Google, including AI Overviews and AI Mode, the file does nothing. Independent studies agree: server-log analyses find AI crawlers fetching robots.txt hundreds of times for every fetch of llms.txt, and a 300,000-domain study found no relationship between having the file and being cited.
What Google Chrome Says
In May 2026 the Chrome team added an Agentic Browsing category to Lighthouse, the audit tool built into Chrome DevTools. It measures how ready a page is for AI agents that browse on a person's behalf, rather than for people or for search crawlers. One of its audits is called llms.txt.
Here is Google's own documentation for that audit, quoted in full. It is three paragraphs long.
The llms.txt file is an emerging convention used to provide a machine-readable summary of a website's content, specifically designed for LLMs and AI agents.
Without this file, agents may spend more time crawling the site to understand its high-level structure and primary content.
How the llms.txt audit works
Lighthouse flags the pages if a server error occurs when attempting to retrieve the llms.txt file. If the file is not provided by the server (resulting in a 404), the audit is marked as Not Applicable (N/A), as providing the file is optional at the moment.
How to fix
Create an llms.txt file and place it in the root directory of your website (for example, https://example.com/llms.txt). The file should follow the llms.txt specification and provide a concise Markdown summary of your site's purpose and key links.
What that means in practice
No file, no penalty
A clean 404 is marked Not Applicable. Google is explicit that the file is optional "at the moment". That phrase is doing work: it leaves the door open to the audit becoming a real pass or fail later.
A broken file is worse than none
A server error on the request, such as a 500 or a timeout, is flagged. Sites that publish the file and then let it rot behind a misconfigured redirect or a dead route do worse in this audit than sites that never bothered.
The category is scored as a ratio, not 0 to 100
Agentic Browsing reports how many of its checks you pass, alongside WebMCP tool registration, accessibility-tree integrity and layout stability. Google calls the category experimental and says it is based on proposed standards. Read the scoring page for the full list.
Why the Two Answers Do Not Conflict
It looks like Google contradicting itself. It is not. Two teams are answering two different questions, and the difference is the whole point.
Google Search has crawled your site for years and trusts its own index over anything you tell it about yourself. An AI agent sent to your site by a user has none of that. It lands on the homepage and has to decide what to fetch next. The Lighthouse sentence quoted above is a statement about that second reader, and it is a claim you can test.
So Do You Need an llms.txt File Today?
We tested the Lighthouse claim rather than taking either Google team's word for it. Our AI Read Cost benchmark asks an agent the same questions about a site with and without its llms.txt and counts the tokens it burns getting to a correct answer. The answer depends on one thing.
When the file is under roughly 10,000 tokens. The agent reads the index, picks the right page and stops. Google's "may spend more time crawling" holds.
When the file is over that size. The agent swallows a dump of your whole site before it can do anything. A big llms.txt is worse than no file at all.
Lighthouse does not check size. It only checks that the request does not error. Google's own "how to fix" says the file should be "concise", and our data says concise is the difference between helping and hurting. So the honest decision tree is short.
You want to rank in Google, including AI Overviews
Skip it. Google has told you it does not read the file, and no independent study has found it moves citations in any AI search engine. Spend the time on the things Google does list: clear text content, working structured data and pages that render without JavaScript.
AI agents or coding tools read your site directly
Publish one, and keep it small. Documentation sites, API references, product catalogues and anything a Claude or ChatGPT agent might be sent to on a user's behalf benefit measurably. This is the reader Lighthouse is auditing for.
You already have one
Check that it still loads and that it is an index, not a dump. A file that errors fails the Lighthouse audit. A file that has grown past a few thousand words costs every agent that reads it. Both are one free audit away from being found.
Where the web is: across a ranked sample of the top 20,000 sites, our registry finds fewer than one in ten publishing an llms.txt. The file is optional, most sites treat it that way, and Google's "at the moment" is the only signal that this might change.
What Lighthouse Checks, and What It Leaves Out
The Lighthouse audit is a reachability test. It answers "did the request error?" and nothing else. Our AI Readiness Audit runs the same test and then asks the questions that decide whether the file helps.
Each check links to its own page, which documents exactly what the code tests and how to fix a failure.
Find Out What an Agent Sees
Lighthouse tells you whether the request errored. We tell you whether the file is reachable, valid, small enough to help, and whether the rest of your site is ready for the agent that reads it.