On 8 June 2026, Apple rewrote its "About Applebot" documentation to formalise that content crawled from publisher sites may be used to train Apple's AI foundation models and to generate answers inside Siri. The revision landed on the same day Apple announced Siri AI at WWDC - a rebuilt assistant that handles a reported 1.5 billion daily requests and can now answer virtually any question from the web without the user opening a browser. For publishers who have already adapted to Google AI Overviews and ChatGPT, Siri AI is the third major AI answer surface that consumes their content and, for now, returns nothing.
What Apple announced at WWDC 2026
Apple introduced Siri AI on 8 June 2026, describing it as a profoundly more capable and conversational assistant with broad world knowledge, personal context understanding, and onscreen awareness. The rebuilt assistant can answer questions from the web on virtually any topic, continue conversations with follow-up questions, and surface information from a user's messages, email, and photos.
The underlying architecture uses a three-tier privacy model. Simple tasks run on-device using Apple's own foundation models. Moderate tasks route to Apple's Private Cloud Compute. Heavy reasoning tasks only route to Google Cloud, where Google's Gemini models handle the request. Apple's World Knowledge database - the index of web content that Siri draws on to generate answers - is Apple's own, built from content crawled by Applebot.
On the same day, Apple published a significant rewrite of its "About Applebot" documentation, formalising for the first time that data crawled by Applebot "may be used to help train Apple foundation models powering generative AI features across Apple products, including Apple Intelligence, Services, and Developer Tools." Applebot previously served Spotlight, Siri suggestions, and Safari search. It now explicitly feeds an AI training pipeline and a real-time answer generation system.
Why publishers need to pay attention
Siri AI is the third major AI answer layer publishers need to plan for. The other two - Google AI Mode and ChatGPT - have been covered extensively. Siri AI differs in one structurally important way: it can fire before the user opens a browser at all. A person asking Siri a question on a locked iPhone screen, on AirPods while commuting, or via HomePod never reaches the browser layer. The question is answered at the device level, the content is consumed, and the publisher earns nothing.
Apple has 2.2 billion active devices. Siri is present on every one of them. Chartbeat data from early 2026 shows that small publishers have already lost 60% of their search referral traffic over two years as AI answer layers expanded. Siri AI adds a parallel zero-click surface across the most widely used mobile operating system in the world. A publisher that has adapted its content strategy for Google AI Mode but has not thought about Siri faces a compounding gap.
Apple has not disclosed how many citations Siri AI sends back to publishers, what traffic those citations generate, or whether source links are consistently shown to users. The documentation states that Siri answers "may include links to sources and websites used to help generate the answer," but there is no Search Console equivalent for Siri. Publishers currently have no first-party data on their Siri citation rate.
What Applebot is now doing on your site
The June 8 documentation update formalises two distinct uses of Applebot-crawled content, and publishers need to understand the difference between them.
The first use is AI training. Apple states explicitly that crawled data may help train its foundation models - the models that power Apple Intelligence and Siri across the product line. Training use is separate from any individual answer generation: content crawled today may influence model behaviour months later.
The second use is real-time retrieval. Apple's documentation describes a retrieval-augmented generation system in which crawled content "may provide additional context and up-to-date content when AI models are used to generate output for display in Apple products and services." This is the mechanism behind Siri answering a world-knowledge question and potentially citing the source.
These two uses are governed by different technical controls. Publishers who assume that addressing one handles both are almost certainly wrong.
How publishers can control Applebot
Apple provides three independent controls. Understanding what each one does and does not affect is essential before making changes.
Applebot-Extended is a secondary user agent that does not crawl independently. It evaluates content already fetched by the primary Applebot to determine its eligibility for AI training. Publishers who want to prevent their content from being used to train Apple's models add a disallow directive for Applebot-Extended in robots.txt. Critically, this has no effect on search indexing: pages that disallow Applebot-Extended can still appear in Spotlight, Siri suggestions, and Safari results.
The nosnippet meta tag controls real-time AI retrieval. Apple will not use content marked with nosnippet as context when generating AI answers in Siri. Publishers can apply this at the page level in HTML or use the Applebot-specific variant: <meta name="applebot" content="nosnippet">. Applying nosnippet also prevents Applebot from generating descriptions or web answers for the page in conventional Apple search results - a trade-off to weigh before applying it broadly.
The isAccessibleForFree schema.org property handles subscription content. Pages marked with "isAccessibleForFree": false in JSON-LD structured data remain eligible for search indexing, but Apple will not use paywalled text as context in AI-generated answers. This is the most targeted control for publishers operating metered or subscriber-only models.
One notable technical detail from the updated documentation: Applebot does not follow the crawl-delay directive. Publishers who rely on crawl-delay to manage Apple's server load should not assume it is being honoured.
What this means for content strategy
The first practical step is audit, not alarm. Publishers should check their robots.txt files to understand their current Applebot-Extended stance. Without an explicit directive, Apple's default is to allow training use. Most publishers inherited this default without making a deliberate choice about it.
Beyond the defensive posture, Siri AI creates a genuine citability opportunity. When Siri answers a world-knowledge question and includes source links, the sites cited get brand exposure on a device surface that has never carried this kind of AI citation before. The conditions for being cited by Siri are the same conditions that improve citability across all major AI systems: direct answers, verifiable facts, clear structure, question-format headings, and authoritative sourcing. Optimising for Siri AI is not a separate content task - it is the same body of GEO work that already improves visibility in Google AI Mode, ChatGPT, and Perplexity.
The deeper issue for publishers is one that Siri AI sharpens but did not create. When content is read at the AI layer - whether by Google's crawlers, OpenAI's retrievers, or Applebot - the read is invisible to client-side analytics. No GA4 session fires. No ad impression is recorded. No revenue is earned. The visit that would have generated an impression and a CPM never happens. Siri AI adds one more surface on which publisher content is consumed before a browser opens, at a scale that makes the existing measurement and monetisation gap considerably harder to ignore. The question of how that gap closes remains, for now, unanswered by any of the major AI platforms.
Frequently asked questions
What is Siri AI and when did Apple launch it?
Siri AI is Apple's rebuilt AI assistant, announced at WWDC 2026 on 8 June. It replaces the previous rule-based Siri with a conversational assistant capable of answering web questions on virtually any topic, continuing dialogue with follow-up questions, and understanding personal context from a user's messages and emails. The assistant uses Apple's own foundation models for most tasks and Google's Gemini models for heavy reasoning tasks only. Apple's World Knowledge database - the content index the assistant draws on - is Apple's own.
Does Siri AI send traffic back to publishers?
Apple's documentation states that Siri answers "may include links to sources and websites used to help generate the answer." However, Apple does not disclose how frequently citations appear, what traffic volumes they generate, or whether source links are consistently visible in the interface. There is currently no publisher-facing reporting for Siri AI equivalent to Google Search Console. Publishers have no first-party data on how often Siri cites them.
What is Applebot-Extended and how is it different from Applebot?
Applebot is Apple's primary web crawler, responsible for indexing content for Spotlight, Siri, and Safari search. Applebot-Extended is a secondary agent that does not crawl independently - it evaluates content already fetched by Applebot to determine whether it can be used to train Apple's AI foundation models. Disallowing Applebot-Extended in robots.txt blocks training use only; it does not remove a site from Apple search results.
Can publishers opt out of Siri AI content use entirely?
Publishers can block two uses separately: AI training (via an Applebot-Extended disallow rule in robots.txt) and real-time AI citation (via the nosnippet meta tag). Applying both leaves the site in Apple's search index but prevents Apple from using its content for model training or answer generation. To stop Applebot from crawling at all, publishers must disallow the primary Applebot user agent - which removes the site from Apple search entirely.
How does Siri AI affect publisher revenue?
Siri AI adds a zero-click surface on which publisher content is consumed without generating a page view or ad impression. Apple has not introduced a compensation model for publishers whose content Siri summarises. The revenue impact mirrors the pattern already established by Google AI Overviews and ChatGPT: content value is transferred to the AI platform and the publisher earns nothing from that specific interaction. Chartbeat data from 2026 shows small publishers have already lost 60% of their search referral traffic over two years as AI answer layers expanded. Siri AI adds another channel to that compounding loss across Apple's 2.2 billion active devices.