From February 1, 2027, Bing reads AI training and AI grounding as separate settings. Google announced a plan for 2027 with no syntax yet. Every date, who must act, and the exact lines from Bing's post.
Client result: Wearview went from $0 to $36K MRR in 8 months.
This post is part of our Technical SEO guide. Related reading: How to slow down Googlebot and AI crawlers with 429 and, Cloudflare's September 15 AI Crawler Change, Your robots.txt GPTBot Block Probably Isn't.
Abdelmoghit drives the content engine at AY Rank. He researches keywords, plans content clusters, and produces citation-optimized articles that rank in both Google and AI search engines.
Full Bio →AI training and AI grounding are now two separate settings at Bing. On October 30, 2026, Google and Microsoft both published posts about AI controls for websites. Google announced a plan. Bing shipped syntax, a deadline and a grounding control. This page puts both posts side by side, quotes them, and tells you who needs to do what and when. If your site has never used NoArchive or NoCache, the short version is that nothing changes for you unless you want it to.
Bing's post defines both terms. Google's post only talks about training.
Bing adds one line that matters for SEO teams: "Indexing and ranking pages in traditional search results is not grounding." So neither setting changes your eligibility for Bing's blue links. Bing says it directly: "These usage preferences do not change eligibility for traditional Bing search results, subject to normal crawl and indexing requirements."
Google changed nothing yet. It announced a plan. The post, by John Mueller on October 30, 2026, says Google will support "a shared AI training control in robots.txt" based on work from the IETF AIPREF working group.
Here is what the post says, and what it leaves out.
| Question | What Google says |
|---|---|
| What is coming? | A shared "policy token" in robots.txt to "set broad preferences across participating AI providers" |
| When? | "We aim to support this new control along with other AI providers in 2027" |
| What if I opt out with the shared token? | If "Google-Extended is not explicitly specified, Google plans to interpret that as if the publisher had disallowed Google-Extended" |
| Does Google-Extended still work? | Yes. "Existing company-specific controls will continue to function alongside this new joint signal" |
| Grounding | "Websites that specifically allow Google-Extended are eligible to appear in responses and receive clicks via grounding in Gemini Apps" |
| Syntax for the shared token | Not given |
| A separate grounding control | Not given |
Google-Extended in robots.txt, which Google calls "the AI training control we've offered since 2023", is the Google control to use. Any article that hands you a "Google" line for the shared token is guessing.
Bing's post links an IETF Internet-Draft that defines the token __usage_ai-training. Google's post does not name that draft or that token. It may end up the same. Google has not said so.
Bing shipped two separate settings and a transition plan. The post, by Krishna Madhavan of Microsoft Bing, says the new preferences "will replace existing Microsoft preferences NoArchive and NoCache for training and grounding."
| AI Training | AI Grounding | |
|---|---|---|
| Where you set it | robots.txt (site or path), or a single page | A single page only. "There is no site-wide robots.txt grounding preference" |
| Values | Allow, Disallow | Allow, DisallowExtended, Disallow |
| Default | Allow | Allow |
| Page syntax | <meta name="Bingbot" content="ai-training:disallow"> or the X-Robots-Tag header | <meta name="Bingbot" content="ai-grounding:disallow"> or the X-Robots-Tag header |
| Priority | "The new AI Training preference set at a page level overrides the robots.txt setting" | Page level only |
| Temporary Bing Webmaster Tools setting | None | Yesfrom December 15, 2026, only for NoArchive or NoCache sites |
DisallowExtended is the new middle option. Bing says it "limits grounding use to Microsoft-operated experiences such as AI Mode, AI Overview, Copilot, and Microsoft 365, and excludes grounding services Microsoft provides to third parties, such as Web IQ and grounding through API access in Foundry."
Bing is blunt about the cost: "If you select DisallowExtended or Disallow for grounding, you will see reduced brand visibility in AI experiences." It also says "Neither choice guarantees inclusion, traffic, or payment."
To check whether your grounding choice changes anything, look at Bing's own data. The AI Performance report in Bing Webmaster Tools counts how often Copilot and Bing's AI summaries cite your pages, and shows a sample of the grounding queries behind those citations. For citation counts in ChatGPT, Perplexity and Gemini you need a third-party tool, compared in our guide to AI citation monitoring tools.
Since October 13, 2026, that report also has a Topics Portfolio tab (beta) that lists queries where other sites are cited and yours is not. Before you write a page for one of them, check the page isn't opted out of grounding. Our guide to turning Bing's Topics Portfolio gaps into pages has the full checklist.
Here is every date in the two posts. Three of them apply only to sites that use NoArchive or NoCache today.
| Date | What happens | Who it affects | Source |
|---|---|---|---|
| October 30, 2026 | Google announces its plan. Bing announces AI Training and AI Grounding | Nobody has to act | Google, Bing |
| December 15, 2026 | Temporary grounding transition preferences open in Bing Webmaster Tools | Only sites using NoArchive or NoCache as of that date. "All other sites need no transition action" | Bing |
| February 1, 2027 | "Microsoft starts honoring AI Training and AI Grounding preferences as pages are crawled" | Any site that publishes the new preferences | Bing |
| March 1, 2027 | After this date, "NoArchive and NoCache will no longer be read as a preference for training or grounding" | Sites using NoArchive or NoCache | Bing |
| September 30, 2027 | Temporary Bing Webmaster Tools grounding preferences expire, or earlier, "until Microsoft recrawls the page and discovers its new AI Grounding preference" | Sites that used the temporary setting | Bing |
| 2027, no date given | Google aims to support the shared training control "along with other AI providers" | Everyone, once Google publishes details |
Most sites need to do nothing. Find your row.
| Your site today | Bing AI Training | Bing AI Grounding | |
|---|---|---|---|
| No NoArchive or NoCache, defaults are fine | Nothing. Default is Allow | Nothing. Default is Allow | Nothing |
| No NoArchive or NoCache, you want to opt out of training | Add a __usage_ai-training group to robots.txt, or a page meta tag or header. Bing honors it as it crawls from February 1, 2027 | Nothing, unless you also want to limit grounding | Google-Extended in robots.txt works today. Google aims to support the shared control in 2027 |
| You want to limit grounding on some pages | No change needed | Add ai-grounding:disallow or ai-grounding:disallowextended to each page. There is no robots.txt option | No grounding control in Google's post |
| You use NoArchive or NoCache today | Publish your training preference in robots.txt or on pages before March 1, 2027. There is no temporary training setting | If you want to keep limiting grounding: from December 15, 2026, record a temporary grounding preference in Bing Webmaster Tools, and publish page-level grounding preferences. Both before March 1, 2027 | Google-Extended works today |
If you use NoArchive or NoCache and skip the Bing Webmaster Tools step, Bing says your pages "default to Allow for grounding until Microsoft recrawls them and discovers your new AI Grounding preference." Bing asks for the temporary setting "because we may not retrieve your new preferences until we recrawl your site; an activity that may not take place between February and March 1."
These examples are copied word for word from Bing's post. Change nothing but the path.
# Allow is default. No action needed to allow ai training
# Disallow ai training for whole site
User-Agent: __usage_ai-training
Disallow: /
# Disallow ai training at a path-level
User-Agent: __usage_ai-training
Disallow: /example-path/
Bing's optional explicit Allow, also word for word:
# Allow is default. No action needed to allow ai training. To specify explicity use
User-Agent: __usage_ai-training
Allow: /
Use either the meta tag or the HTTP header:
<meta name="Bingbot" content="ai-training:disallow">
X-Robots-Tag: Bingbot: ai-training:disallow
<meta name="Bingbot" content="ai-grounding:disallow">
X-Robots-Tag: Bingbot: ai-grounding:disallow
For the middle option, Bing's post says: "Use ai-grounding:disallowextended for the other grounding preferences." That gives content="ai-grounding:disallowextended" in the same meta tag, or Bingbot: ai-grounding:disallowextended in the header.
User-agent: * does not count. The IETF draft that Bing links says "wildcard groups (i.e., User-Agent: *) MUST NOT be interpreted as matching the policy token." A Disallow: / under User-agent: * blocks crawling. It does not express an AI training preference. You need a group that names __usage_ai-training.
Our robots.txt generator builds the crawler rules. It does not write the __usage_ai-training group yet, so paste that group in by hand. Check what a page actually says with our free AI training and grounding checker. It reads the robots.txt, meta tags and X-Robots-Tag header and shows the result for each setting. If you have set up AI crawler rules before, our robots.txt setup guide for AI crawlers covers GPTBot, ClaudeBot and the rest.
No. These preferences govern use, not crawl access. Bing says "These settings govern use of content, not crawl access. An Allow setting does not override a robots.txt rule blocking Bingbot from crawling." The IETF draft says the same about the token: "This policy token is not a directive for crawling behavior."
robots.txt is a request that crawlers choose to follow, not a lock. We cover what that means for AI crawlers in why blocking GPTBot in robots.txt often doesn't work. The same habit helps here: check what each line says for the exact path, not what you meant it to say.
Two more limits from Bing's post:
That is a business call. Bing names a visibility cost only for grounding. On training, its one line on the upside is "Allowing training can help improve future generative models." Blocking grounding costs you visibility. Bing calls grounding the path that brings "relevant sources and web publishers' expertise into the response," with "links and citations providing a path to the original content."
Our view: if you want to be cited in AI answers, leave grounding on. Decide training on its own terms. That is the split Bing's own example shows: "you can set AI Training to Disallow and AI Grounding to Allow to keep content eligible to inform generated responses without allowing it to train generative models."
If you sell on being found in AI search, grounding is the setting to protect. A technical SEO audit is the place to check that your crawl rules, meta tags and headers say what you mean, page by page. If you would rather walk through it with us, book a call.
Training uses your content to change what a future AI model learns. Grounding uses your page as input when an AI answer is created, with "links and citations providing a path to the original content." Bing lets you set the two separately, and both default to Allow.
No. December 15 is when Bing Webmaster Tools opens a temporary grounding setting, and only for sites using NoArchive or NoCache. Their deadline is March 1, 2027. Bing says "All other sites need no transition action."
No. Bing says "There is no site-wide robots.txt grounding preference." You set grounding page by page, with <meta name="Bingbot" content="ai-grounding:disallow"> or the X-Robots-Tag header.
__usage_ai-training today?Google has not said so. Its October 30, 2026 post announces a plan to support a shared AI training control in 2027 and gives no syntax. Google-Extended remains Google's working training control in robots.txt.
Bing says no. Its post says these preferences "do not change eligibility for traditional Bing search results, subject to normal crawl and indexing requirements." Blocking Bingbot from crawling is a different rule, and that one does affect search.
It is a Bing grounding value that limits grounding to Microsoft's own AI experiences, such as Copilot, AI Overview, AI Mode and Microsoft 365. It excludes grounding services Microsoft provides to third parties, such as Web IQ and the grounding API in Foundry. Bing adds that no choice guarantees inclusion.
User-agent: * with Disallow: / opt me out of AI training?No. The IETF draft linked from Bing's post says wildcard groups "MUST NOT be interpreted as matching the policy token." That rule blocks crawling for every bot, which is a much bigger change. Use a User-Agent: __usage_ai-training group instead.
Sources: Google Search Central Blog, Supporting a shared AI training control for websites (October 30, 2026), Bing Search Blog, Your content, your choice: clearer preferences for training and grounding (October 30, 2026), IETF Internet-Draft draft-kelley-aipref-training-preference-signal-00, Bing Webmaster Tools FAQ, New AI Preferences, Bing Webmaster Blog, Introducing AI Performance in Bing Webmaster Tools Public Preview (February 10, 2026)