In short. Microsoft documents nocache as limiting Chat and Copilot display and foundation-model training to a page's URL, title and snippet.

Definition: nocache is a robots meta directive that Microsoft uses to restrict content use in its AI answers. Bing's publisher-control announcement of 22 September 2023 specifies that Bing Chat may display only a tagged page's URL, title and snippet, the short description accompanying a search result.

Key facts

As of 17 October 2026, Microsoft's robots meta tag reference retains nocache for Chat and Copilot and limits model training to URLs, titles and snippets.
Bing's original announcement states that pages carrying nocache or noarchive still appear in search results.
The current Bing Webmaster Guidelines describe noarchive as preventing content use in Copilot responses and grounding results. Grounding means supplying retrieved information to support an answer.
Google's robots meta tag specification lists nocache among unused rules and states that Google Search does not use it.

What does Bing's original announcement establish?

The Bing Chat announcement distinguishes answer display from future training use. For indexed content labelled nocache, both are limited to the URL, title and snippet. For content labelled noarchive, it excludes inclusion and links in Bing Chat answers and future use in Microsoft's foundation-model training.

It also specifies an unusual combination rule: a page carrying both directives is treated as nocache. Adding that value to noarchive therefore permits the limited use described for nocache.

What do Microsoft's current documents say?

The current meta tag reference continues to describe cached-page access alongside AI content use. Its nocache entry instructs Bing to withhold a cached page, display only the URL, snippet and title in Chat or Copilot, and restrict foundation-model training to those elements. It repeats the combination rule that gives nocache precedence over noarchive.

The current Webmaster Guidelines confirm the Copilot limit. They separately describe noindex as the instruction for excluding a URL from Bing search, Copilot experiences and grounding API results. These distinctions matter when choosing between retaining search visibility, limiting answer content and excluding the URL.

How is nocache implemented and checked?

Microsoft's meta tag reference provides this page-level form:

<meta name="robots" content="nocache">

The same reference allows name="bingbot" to target Bing specifically:

<meta name="bingbot" content="nocache">

It also permits the directives in an HTTP response header, a field returned by the web server:

X-Robots-Tag: nocache

An implementation review checks the delivered HTML and response headers for the intended value and scope. It also checks for an existing noarchive value because combining the directives changes Microsoft's stated treatment. A plugin setting should be checked against the actual page response.

Google's specification explains that page directives require crawler access to be read and followed. Blocking the page through robots.txt, the file containing crawl-access rules, prevents that reading. Crawl access therefore belongs in the review alongside the meta tag and headers.

FAQ

Does nocache remove my page from Bing search?

Bing's announcement explicitly retains search-result appearance for pages carrying nocache or noarchive.

Can I combine nocache and noarchive for a stronger restriction?

Microsoft's reference says that the combination is treated as nocache. If you intend the exclusion described for noarchive, adding nocache changes that instruction to the limited-use option.

Does nocache control Google Search?

Google's specification states that Google Search does not use this rule. Its documented effects must be assessed separately for each engine.

Sources

Check how a page reads to answer engines. Open the GEO score analyzer