TL;DR
- Apple updated its Applebot documentation over the weekend of 7 September 2026 to state that site rules for Applebot-Extended are not considered in ranking for Search.
- The same page states that Applebot-Extended does not crawl webpages and that pages disallowing it can still be included in search results.
- Applebot-Extended only governs whether data already crawled by Applebot may train Apple's generative foundation models; the opt-out is a robots.txt user-agent line plus a disallow rule.
- Barry Schwartz reported the documentation change at Search Engine Roundtable on 7 September 2026; it mirrors Google's split of Google-Extended from Googlebot.
- The statement covers Apple Search only, says nothing about Google, OpenAI, Anthropic or Perplexity crawlers, and is documentation rather than an independent test.
Apple has stated in its own documentation that blocking Applebot-Extended does not affect a site's position in Apple Search. The Applebot support page was updated over the weekend of 7 September 2026 to add an explicit line: "Site rules for Applebot-Extended are not considered in ranking for Search."
Applebot-Extended is Apple's secondary user agent, and its only job is to let publishers control whether their content is used to train Apple's generative foundation models. Barry Schwartz spotted the addition and reported it at Search Engine Roundtable on 7 September 2026. The wording removes the specific fear that has kept a lot of publishers from writing the opt-out rule at all.
What Apple added to the Applebot documentation on 7 September 2026
The change is one sentence, and the sentence is the whole story. Apple's documentation now says: "Site rules for Applebot-Extended are not considered in ranking for Search." Before that line existed, a site owner writing an opt-out rule had to decide on faith whether Apple treated the block as a signal about the site. Now the answer is written down by the company that runs the crawler.
The same document carries three other statements that matter as much as the ranking line. It says "Applebot-Extended does not crawl webpages." It says "Webpages that disallow Applebot-Extended can still be included in search results." And it says "Applebot-Extended is only used to determine how to use the data crawled by the Applebot user agent." Read together, those sentences describe an agent that is a permission flag rather than a crawler.
Two user agents doing two different jobs
Applebot is the crawler. It fetches webpages, and its index is what sits behind Apple Search across Siri, Spotlight and Safari suggestions. If a site blocks Applebot, it is blocking the thing that finds and reads its pages.
Applebot-Extended is not a second crawler. Apple's documentation states that it does not crawl webpages, and that it is only used to determine how to use the data that Applebot has already collected. In practice it is a switch applied after the fetch, and the switch governs one thing: whether the content that Applebot brought back may be used to train Apple's generative foundation models.
Once that separation is clear, the ranking sentence stops being surprising. A rule aimed at an agent that never crawls anything is a rule about downstream usage rights, so there is nothing in it for a ranking system to read. Apple has now said as much rather than leaving site owners to infer it.
What the documentation says, line by line
The table below quotes the statements in Apple's Applebot documentation that a publisher needs before making a decision, with nothing added.
| User agent or rule | What Apple's documentation says |
|---|---|
| Applebot | The crawler whose data is at issue. Apple describes Applebot-Extended as governing "how to use the data crawled by the Applebot user agent". |
| Applebot-Extended | "Applebot-Extended does not crawl webpages." It exists so publishers can control whether content is used to train Apple's generative foundation models. |
| A disallow rule for Applebot-Extended | "Webpages that disallow Applebot-Extended can still be included in search results." |
| Effect on Apple Search ranking | "Site rules for Applebot-Extended are not considered in ranking for Search." |
How to write the Applebot-Extended robots.txt rule
The opt-out is a robots.txt block that names the user agent and then applies a disallow rule. The user agent line is written exactly as User-agent: Applebot-Extended, and the disallow line beneath it carries the path being withheld, with a single forward slash covering the whole site.
Three practical points about the file itself. It has to live at the root of the domain, because that is the only place a crawler looks for it. Each subdomain has its own robots.txt, so a rule on the main domain does not travel to a blog or shop hosted on a subdomain. And the file is public, so anyone can read the rule after it is live, including you: open the domain followed by /robots.txt in a browser and confirm the two lines are there and spelled correctly.
Spelling is where these rules fail most often. A user agent token is matched as written, so Applebot-extended or AppleBot-Extended are not safe substitutes for the exact string, and a rule that no agent matches is a rule that does nothing while looking like it works.
How this compares with Google-Extended
The structure will look familiar to anyone who has already handled Google-Extended. Google separated Google-Extended from Googlebot and from Search ranking, and Apple's documentation now describes the same separation for its own stack: one agent that crawls and feeds search, a second token that governs training use and carries no ranking consequence.
The useful part of the parallel is that a publisher who has already made the Google-Extended decision has, in effect, already had the argument internally. The question is the same one: is the content worth more as training material for someone else's model, or withheld. What is new is only that Apple has now put its half of the answer about ranking in writing.
What a site owner actually has to decide
The decision is not technical, because the technical part is two lines in a text file. It is editorial and commercial.
- What the content is worth as training data. A site whose value is a large body of original writing, original photography or proprietary research is giving away something different from a site whose pages are mostly product listings and service descriptions.
- Whether the block is meant to be total or partial. A disallow rule can name a path rather than the whole site, so a publisher can withhold an archive of long-form work while leaving marketing pages open.
- Who owns the file. robots.txt usually sits with engineering while the decision about content rights sits with editorial or legal, and a rule that nobody owns tends to get overwritten in the next deployment.
- What you will do about the other crawlers. Apple's statement covers Apple. Deciding about Applebot-Extended without deciding about the rest is half a policy.
For teams working through where their content shows up in generated answers, this decision sits alongside the wider set of choices covered in AI SEO work, and the visibility side of it belongs with generative engine optimisation. Blocking training use and wanting to be cited in AI answers are separate goals that can be held at the same time, because they are governed by different rules and different agents.
What this statement does not cover
It is Apple's statement about Apple Search. It says nothing about how any other search engine treats an Applebot-Extended rule, and nothing about rankings anywhere else.
It says nothing about Google, OpenAI, Anthropic or Perplexity crawlers. Each of those has its own tokens and its own published behaviour, and a decision about Apple does not transfer to any of them.
It is a documentation statement, not an independently tested result. Nobody outside Apple has run a controlled test showing that blocked sites hold their positions; what exists is Apple's written commitment, which is a reasonable thing to rely on and still a different category of evidence from a measurement.
And it is a statement about ranking, which is a narrower claim than a statement about traffic. Apple's documentation says webpages that disallow Applebot-Extended can still be included in search results and that the rules are not considered in ranking. It does not make any promise about how much traffic a site receives.
What this means for Thai marketers
Apple's documentation does not mention Thailand or any other market, and nothing in the update is region-specific as reported. What changes for a Thai site owner is the same thing that changes for everyone: the ranking fear behind the decision now has a written answer for Apple's stack.
The practical weight of that depends on how much a site relies on Apple surfaces. Thai publishers, media sites and content-heavy brands with a large iPhone audience have a real interest in Siri, Spotlight and Safari suggestions, and those are the sites where the anxiety about blocking has been most reasonable. For a small local business site whose traffic arrives mainly through Google and social, the decision matters less either way.
There is also a rights dimension that Thai publishers producing original Thai-language work should weigh on its own terms. Thai-language content is comparatively scarce as training material, which makes the question of whether to hand it over a genuine decision rather than a formality. Apple's update does not answer that question. It only removes one bad reason for avoiding it.
FAQ on Applebot-Extended and Apple Search
Will blocking Applebot-Extended hurt my Apple Search rankings?
No, according to Apple's own documentation, which states that site rules for Applebot-Extended are not considered in ranking for Search. The same page says webpages that disallow Applebot-Extended can still be included in search results.
Does blocking Applebot-Extended remove my site from Siri or Spotlight?
Apple's documentation says pages that disallow Applebot-Extended can still be included in search results, and it separates Applebot-Extended from the crawler that collects pages. The rule governs training use of already-crawled data rather than indexing.
Does this apply to Google, OpenAI or Perplexity crawlers as well?
No, and the source did not address them. Apple's statement is about Apple Search and the Applebot-Extended token; every other company publishes its own agents and its own behaviour, and each needs a separate decision.
Do I have to do anything right now?
No, this is a documentation update rather than a change that requires action, and sites with no Applebot-Extended rule are in exactly the position they were in last week. The update matters only if the ranking risk was the thing stopping you from writing the rule.
Is this confirmed for Thailand?
The documentation does not break anything out by country, so there is no Thailand-specific statement to cite. It is written as a general description of how Apple treats the token.
If the training-data question has been sitting unanswered on someone's list since the first AI crawlers appeared, this is a reasonable week to close it. Relevant Audience can walk through what your robots.txt currently allows and what the trade-off looks like for your content before anything gets changed.
