Nosnippet and Data-Nosnippet in AI Answers: What They Hide and How to Use Them

Google and Bing document four on-page controls that limit what their AI features can show. Three work on the whole page. One works on a section.

The difference that matters is indexing. noindex takes the page out of search entirely. The snippet controls leave the page in the index and only stop its text from being displayed. For a fuller side-by-side, see noindex vs nosnippet vs data-nosnippet for AI.

max-snippet also accepts a positive number, which Google's robots meta tag documentation defines as a cap on how many characters a snippet may show. A value of -1 means no limit. A value of 0 means no snippet at all. The max-snippet definition covers the accepted values.

What this means: pick the control by the question you are answering. "Should this page exist in search?" is a noindex question. "Should this text be quoted?" is a snippet question.

Google: AI Overviews and AI Mode

Google's AI features documentation states: "To be eligible to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet." It adds that there are no additional technical requirements.

That sentence is the whole mechanism. A page with noindex is not indexed. A page with nosnippet or max-snippet:0 is not eligible for a snippet. Either way, it fails the gate. This is what index eligibility means in practice for Google's AI surfaces.

Eligibility is not inclusion. The same Google page says indexing and serving isn't guaranteed. Passing the gate only means you are not excluded by your own tags.

There is no robots.txt token that blocks AI Overviews or AI Mode on its own. Google treats AI as part of Search, so Googlebot rules and these on-page directives are the controls. Google is also rolling out a Search Console control for generative AI features. As of October 2026 it is opt-out: inclusion is the default for all properties, and child properties inherit the parent's setting by default.

Bing: snippets and AI summaries

Bing added support for data-nosnippet on 15 October 2025. Its announcement says content marked with it "is still indexed normally, but it will be excluded from snippets and AI summaries." The same post says the tagged content "is available for ranking." So Bing removes display, not ranking eligibility. For what this means for Microsoft's assistant, see whether data-nosnippet keeps content out of Copilot answers.

Other engines

The evidence we rely on documents these snippet directives for Google and Bing. For ChatGPT, Perplexity and Claude, the documented publisher control is per-bot robots.txt permission (for example, OAI-SearchBot for ChatGPT search). Do not assume a nosnippet tag changes what those engines quote.

Page-level directives go in a robots meta tag in the <head>, or in an HTTP response header. Section-level hiding uses an attribute on an element.

<meta name="robots" content="noindex">
<meta name="robots" content="nosnippet">
<meta name="robots" content="max-snippet:0">
<meta name="robots" content="max-snippet:160">

X-Robots-Tag: nosnippet

<div data-nosnippet>Text to keep out of snippets and AI summaries</div>

A few rules keep this correct:

• name="robots" applies to all crawlers that honour it. A specific crawler name (for example googlebot) scopes the directive to that crawler only.
• The X-Robots-Tag header does the same job as the meta tag. It is the only option for non-HTML files such as PDFs.
• Google's documentation lists span, div and section as the elements that accept data-nosnippet. Put it on the smallest element that holds the text you want hidden.

The data-nosnippet definition covers edge cases in more detail.

Snippet controls sit in the access stage: they decide whether your text is allowed into an answer at all. That part is fully yours. You can read every directive on your site and change it.

What you do not control is everything after the gate. Google decides whether an AI Overview triggers for a query. The retrieval system decides which eligible pages enter the model's context. The model decides which of those it cites.

So be precise about the direction of effect:

• Removing an accidental nosnippet restores eligibility. It does not promise a citation.
• Adding data-nosnippet reliably hides a section on Google and Bing. It does not change what models learned in training.
• No markup works in reverse. There is no tag that makes a section more likely to be quoted.

One related lever is separate. Google-Extended, a robots.txt token, governs use of your content for Gemini Apps training and grounding. Google documents that it does not affect inclusion in Google Search. Snippet directives and Google-Extended answer different questions, so check both if training use matters to you.

Most sites do not choose to hide content from AI answers. They inherit it. A directive set once in a shared template applies to every page that uses it, and nothing in a dashboard warns you.

The patterns to look for:

• A wrapper that grew. Someone adds data-nosnippet to hide a cookie banner or a legal footer. Later the attribute ends up on a layout component that wraps the main content, and every product description disappears from snippets. See whether a cookie banner can hide your content from AI.
• A staging directive shipped live. A noindex added to keep a staging site out of search survives a deploy. Every page using that layout leaves the index.
• A plugin default. An SEO plugin or CMS setting writes nosnippet or max-snippet:0 on a whole content type, such as all pricing or documentation pages.
• A header nobody reads. An X-Robots-Tag set at the server or CDN layer does not appear in page source. Only the HTTP response shows it.

Say you run a hypothetical SaaS company whose pricing table sits inside a reusable div data-nosnippet block copied from the footer. Google and Bing would still index the pricing page and could still rank it. Neither would quote the prices in an AI summary, which is often the exact fact you most want stated correctly.

Work from the widest control to the narrowest. A page-level block makes section-level checks irrelevant.

1. List what you actually want hidden. Usually a short list: account pages, internal search results, maybe a legal block. Everything else should be eligible.
2. Fetch raw HTTP responses for each key template. Check the X-Robots-Tag header and the robots meta tag on your homepage, product, pricing, category and article templates.
3. Search the HTML for data-nosnippet. Note which element carries it and how much text sits inside. Compare against your list from step 1.
4. Check max-snippet values. A small positive number may truncate the fact you want quoted. Zero removes the snippet.
5. Check the Search Console generative AI control, where it is available to you. Confirm the property and any parent property have not opted out, since children inherit by default.
6. Fix at the template, not the page. Change the shared component so the fix covers every page that uses it.
7. Re-fetch and confirm. Verify the live response, not the CMS setting.

To run this across a whole site, follow the snippet controls site audit playbook. For a quick single-page check, use the free noindex, nosnippet and data-nosnippet check.

• Using noindex when you meant nosnippet. If you only want to stop quoting, noindex removes the page from search entirely.
• Treating data-nosnippet as a ranking penalty. Bing says tagged content stays available for ranking. It hides display only. More on this in whether hiding text from AI summaries hurts rankings.
• Expecting robots.txt to block AI Overviews. No robots.txt token does that. Blocking Googlebot removes you from Google Search as a whole.
• Assuming snippet tags govern ChatGPT or Perplexity. The documented controls there are per-bot robots.txt rules, not these directives.
• Confusing display with training. Snippet directives limit what is shown. Google-Extended governs Gemini training and grounding.
• Adding markup to get quoted. No control in this family is additive. Making a fact quotable means stating it plainly in fetchable HTML, not tagging it.
• Checking the CMS instead of the response. Headers set at the server or CDN never appear in your editor.