Google updated its documentation on controlling content shared through its search engine on October 21, 2026. The page now brings together more explicit explanations of the options website owners can use to manage indexing, the snippets displayed in results and participation in search features powered by artificial intelligence. The change recorded by the company concerns the presentation of existing tools, including robots directives and the control for generative AI features in Search.
The update distinguishes situations with different effects on a website: keeping information outside public access, removing a page from results and restricting how its content is presented. The documentation applies to Google Search. Other Google products, such as shopping listings, hotels and vacation rentals, have their own mechanisms for managing information.
Participation in AI answers has a separate control
In Search Console, the Search generative AI setting allows a website to be included in or excluded from AI Overviews, AI Mode and generative features in Discover. Google says the control was rolled out to websites worldwide starting on August 31, 2026. Exclusion covers links and content used to ground responses in these experiences, without becoming a ranking signal for other Search results.
The setting can be applied to Search Console properties, with an option inherited from a parent property. A child property can have its own selection, and the owner of the main domain is notified of such a change. The control does not change choices made separately for Google Ads or Merchant Center, and does not manage the use of content to train AI models.
For some uses outside search, Google separately documents Google-Extended. This is a product token used in robots.txt through which publishers manage the use of content for future Gemini models and for grounding answers in certain Gemini and Vertex AI products. Google-Extended is not a crawler with its own HTTP user agent and does not influence a website's inclusion or ranking in Google Search. Crawling for search remains associated with Googlebot.
The documentation for AI features retains the same technical condition for a page to appear as a supporting link in AI Overviews or AI Mode: it must be indexed and eligible to appear in Search with a snippet. Relevant factors include crawler access, available textual content, internal links and consistency between structured data and visible information. The two AI experiences can use different models and techniques, so the links presented are not identical across all responses.
Page indexing and snippet display are separate operations
The noindex directive excludes a page from search results after Google crawls and processes it. It can be delivered through a meta element in HTML or through the X-Robots-Tag HTTP header. The header option also allows the rule to be applied to files without HTML, including PDF documents. A page with noindex can remain accessible to people who know its address, because the indexing rule does not introduce authentication or restrict visitor access.
The robots.txt file regulates which addresses crawlers can access. For a web page, blocking crawling can leave the address visible in results if it is discovered through other links. Google can no longer read the blocked content or a noindex directive contained within it. The documentation therefore distinguishes the management of crawler traffic from removing a page from the index; private information is protected through access restrictions on the website itself.
At the preview level, nosnippet stops text snippets and video previews, while still allowing a static thumbnail when one is available. max-snippet sets the maximum snippet length. For part of a page, the data-nosnippet attribute can be applied to span, div or section elements. These controls are also documented as ways to limit the content that can be used directly in AI answers in Search.
Structured data is treated separately: information in markup can be used for rich results even when the corresponding visible area is marked with data-nosnippet. Applying the attribute to page text does not automatically remove data published separately in the page's schema.
For photographs, Google describes blocking through robots.txt or delivering noindex in the HTTP header of the image file. If a header is used, the image address must remain crawlable so the rule can be read. A noimageindex instruction applied to a page concerns images embedded there; the same photograph can be discovered through another page. A rule delivered directly for the file allows the image to be controlled regardless of the page in which it is embedded.
Urgent removals and reporting have their own mechanisms
The Removals tool in Search Console provides a quick intervention to hide an address temporarily. Google indicates a period of approximately six months for this temporary removal. Keeping the content out of results permanently depends on the measure applied at its origin: deleting it, protecting it with a password or adding a noindex directive. The Search Console request and the actual website change consequently have different roles in the removal process.
For AI features, the dedicated Search Console report tracks link impressions in AI Overviews and AI Mode. Data can be analyzed by page, country, device and period; Discover has a separate report. Google generally assigns information to the canonical address and uses the Pacific time zone for reported dates. Totals in the chart and table can differ when aggregation is performed by property and by page respectively.
The revised Search Central page places these options within the same framework for managing content, including on large websites or in sections with substantial user contributions. The October 21 change consists of additional explanations and references to available controls. The effect of each tool remains the one described in its documentation: crawling, indexing, previews, participation in generative features or temporary removal.
