Noindex tag
noindex is a rule that tells search engines not to show a page in their results. At Google, that also keeps it out of AI Overviews and AI Mode.
noindex is a rule that tells search engines not to show a page in their results. You add it to the page as a robots meta tag, <meta name="robots" content="noindex">, or send it in the page's HTTP header as X-Robots-Tag: noindex. It's how you keep thank-you pages, internal search results and old campaign pages out of search.
It matters for AI search because Google's AI features are built on Google Search. A page that's kept out of the index can't be shown as a source in AI Overviews or AI Mode. Google explains the rule in its robots meta tag documentation.
noindex isn't the same as blocking
robots.txt and noindex do different jobs, and they can work against each other.
robots.txt tells a bot which pages it may read. noindex tells a search engine not to list a page it has read. If robots.txt blocks a page, the search engine never reads it, so it never sees the noindex. The page can then still turn up in results, with no description, because other sites link to it.
So if you want a page out of the results, let the bot read it and use noindex.
When it goes wrong
The usual problem with noindex is the opposite one: pages that should be found carry it by mistake. A setting left on after a site redesign, or a plugin that adds it to every new post, can quietly take pages out of search and out of the AI answers built on it.
For example, say Harrow & Pine launches a new range of furniture, and its product pages are copied from a staging site that had noindex switched on. The pages look fine to visitors, but Google won't list them, and AI Overviews won't cite them. Nothing on the page tells you, unless you look at the code or at Search Console.
If an important page stops showing up, check for noindex before anything else.
Other AI platforms
noindex is a search engine rule. AI platforms that run their own search bots follow the rules their companies publish, and those are mostly about robots.txt. Check each company's own page about its bots before you rely on noindex to keep a page out of its answers.