Shahid Ali.WhatsApp

field note

Robots noarchive: Google ignores it, Bing uses it for AI

Crawling and robots.txt

Shahid AliSeptember 7, 2026all posts

Robots noarchive: Google ignores it, Bing uses it for AI

Almost every page explaining noarchive still tells you it controls the Cached link in Google’s results. That link was removed from Google Search in 2024, and Google has since moved the rule into the list of directives it ignores.

So the tag is dead, and everybody can stop writing about it? No. Microsoft gave noarchive a second job that has nothing to do with caching, and if you have the tag sitting on old pages you are running a Bing setting you probably did not choose.

Does noarchive still do anything at Google?

No. Google’s robots meta tag specification files it under the rules that are not used, with a sentence explaining why: “The noarchive rule is no longer used by Google Search to control whether a cached link is shown in search results, as the cached link feature no longer exists.”

That is Google confirming both halves at once. The feature is gone, and the rule that governed it is inert. Leaving noarchive in your HTML does nothing at Google and costs nothing at Google. It is not an error to fix, it is a line that no longer means anything on that side of the fence.

What Bing did with it instead

In September 2023 Microsoft published new controls for how site content is used by its chat and AI systems, and reused two old snippet directives to carry them. Content tagged NOARCHIVE, in Microsoft’s wording, “will not be included in Bing Chat answers and will not be used for training Microsoft’s generative AI foundation models.” NOCACHE was assigned the narrower job, allowing URLs, titles and snippets only.

That is a real, documented, currently live behaviour, and it is the opposite of a deprecation. The same eight characters mean nothing at one search engine and mean an AI opt-out at the other.

One directive, two histories Apr 2020 Sep 2023 2024 2026

Bing 2020: noarchive and nocache published as snippet and caching controls. 2023: same directives, new job. noarchive content “will not be included in Bing Chat answers and will not be used for training Microsoft’s generative AI foundation models”. Status in 2026: live. An AI opt-out.

Google 2024: the cached link feature is removed from search results. 2026: noarchive sits under the rules that are not used. “The noarchive rule is no longer used by Google Search … as the cached link feature no longer exists.” Status in 2026: inert. Ignored.

Dates and quoted wording from the Bing Webmaster Blog, April 2020 and September 2023, and Google Search Central, "Robots meta tag, data-nosnippet, and X-Robots-Tag specifications".

Does noarchive stop AI Overviews?

It does not, and this is the mistake worth catching, because people reach for noarchive precisely when they want out of Google’s AI answers.

Google ignores the rule entirely, so it cannot control anything, AI Overviews included. The directives Google actually documents for this are different ones. On nosnippet, the specification says it “applies to all forms of search results (at Google: web search, Google Images, Discover, AI Overviews, AI Mode) and will also prevent the content from being used as a direct input for AI Overviews and AI Mode.” Google’s page on AI features lists the same short set of levers: nosnippet, data-nosnippet, max-snippet and noindex.

There is a price attached and nobody mentions it. nosnippet does not have an AI-only setting. Switching it on takes you out of AI Overviews and takes away your ordinary search snippet at the same time, so your normal blue link loses its description too. If you only want one section held back, data-nosnippet on that element is the surgical version, and max-snippet:[n] caps the length rather than removing it. I have written the longer version of this decision in what actually keeps content out of AI Overviews.

While we are clearing out wrong answers to the same question: Google-Extended does not remove you from AI Overviews either. It governs a different set of uses, and AI Overviews are grounded in the regular Googlebot index.

noarchive, noindex and nosnippet: which do you want?

Three rules, three completely different outcomes, and the failure mode is always picking the wrong one for the goal.

  • noindex removes the page from results. Total, blunt, and it is what you want when the page should not be findable at all. It cannot be combined with a robots.txt block, since Google has to read the page to see the rule. That trade is covered in noindex versus blocking the crawler.
  • nosnippet keeps the page in results and takes away the description, and at Google that includes use as direct input to AI Overviews and AI Mode.
  • noarchive does nothing at Google, and at Bing removes you from Chat answers and generative AI training.

Read that list against the reason you came here. “I do not want Google’s AI using my content” points at nosnippet or data-nosnippet. “I do not want Microsoft training on my content” points at noarchive. “I do not want this page in search” points at noindex. Almost nobody wants all three.

How do I check the directive is actually being sent?

This is where these rules quietly fail, because a directive that never reaches the crawler is a directive that does nothing, and nothing in your CMS will tell you.

What I run, in order:

  1. Request the page as a plain client and as Googlebot, and compare. curl -sI https://example.com/page/ and then the same with -A "Googlebot". Servers, CDNs and firewalls hand different headers to different user agents more often than people expect, and an X-Robots-Tag set at the edge can appear for one and not the other. The header route is worth understanding on its own terms: see serving the directive as an HTTP header.
  2. Look at the crawled HTML in Search Console. URL Inspection shows you the HTML Google actually fetched, which is the only view that settles an argument about what Googlebot saw. A directive present in your source and absent there is a rendering or edge-caching problem, not a tagging problem.
  3. Check for a meta tag and a header disagreeing. If both exist and conflict, the restrictive one wins, and the one you forgot is usually the restrictive one.

Should I remove noarchive from my site?

Decide it deliberately rather than as cleanup, because the tag is not neutral any more.

If you find noarchive on a site you inherited, it is almost certainly a leftover from when it controlled Google’s cached copy. On the Google side, leaving it costs nothing. On the Bing side, it is currently holding those pages out of Bing Chat answers and out of Microsoft’s generative AI training. If that is what the business wants, keep it and write down why. If nobody knows it is there, it is a decision being made by a previous consultant’s habit.

The place I find these is the per-page advanced settings in the SEO plugins, where a noarchive box was ticked years ago and nobody has looked since. Worth an afternoon on any site that predates 2024.

One thing noarchive has never done, in either engine: it does not affect the Internet Archive. That is a different crawler with its own rules, and no robots meta tag reaches it.

Sources

Message me on WhatsAppRead the guides