Short answer: Microsoft split two choices that used to blur together. Training (can your content change a model’s parameters) can be set in robots.txt. Grounding (can your content inform a live AI answer in Bing, Copilot or Microsoft 365) can only be set page by page, with a meta tag or an HTTP header. Both default to Allow, and Microsoft warns that restricting grounding reduces your visibility in AI answers. The announcement is Bing’s own post of October 30, 2026.

What the two preferences do

  • AI Training is the shared robots.txt preference (User-Agent: __usage_ai-training, covered in our note on the training token). Microsoft also lets a single page override it with <meta name="Bingbot" content="ai-training:disallow"> or X-Robots-Tag: Bingbot: ai-training:disallow.
  • AI Grounding is a Microsoft-specific, page-level preference. Microsoft’s post states there is no site-wide robots.txt grounding preference.
  • Grounding has three states: Allow (default), DisallowExtended, and Disallow.
  • DisallowExtended keeps your content in Microsoft-operated experiences (the post names AI Mode, AI Overview, Copilot and Microsoft 365) but excludes grounding services Microsoft provides to third parties, such as Web IQ and the grounding API in Foundry.

The page-level syntax is <meta name="Bingbot" content="ai-grounding:disallow"> or X-Robots-Tag: Bingbot: ai-grounding:disallow. Use ai-grounding:disallowextended for the middle option.

The trade-off Microsoft states itself

Microsoft writes that choosing DisallowExtended or Disallow for grounding “will significantly reduce your content’s visibility in AI-generated experiences, without affecting eligibility for Bing search results.” Neither choice guarantees inclusion, traffic or payment.

For a store, that is a dial between control and being named in an answer. A product page that carries ai-grounding:disallow can still rank in ordinary Bing results, but it is less likely to be quoted when a shopper asks an assistant what to buy.

Dates that matter

Date What happens
December 15, 2026 Temporary grounding preferences open in Bing Webmaster Tools, only for sites already using NoArchive or NoCache
February 1, 2027 Microsoft starts honouring the new preferences as it crawls
March 1, 2027 NoArchive and NoCache stop counting as a training or grounding preference
September 30, 2027 Temporary grounding preferences expire

If your store uses NoArchive or NoCache today and wants to keep content out of grounding, Microsoft says you must publish the page-level grounding preference and record a temporary preference in Bing Webmaster Tools before March 1, 2027. Otherwise those pages default to Allow for grounding until Microsoft recrawls them.

Two limits worth knowing

  • These settings govern use, not access. An Allow does not override a robots.txt rule that blocks Bingbot from crawling.
  • A published setting only counts once Microsoft recrawls the page. The post notes that IndexNow can raise the chance of a recrawl but guarantees no date.

What to check on your store

  1. Search your templates and CDN rules for noarchive, nocache and any X-Robots-Tag header. Shopify themes, SEO apps and edge rules are the usual places they hide.
  2. Decide per page type. Product and policy pages are what an assistant cites when it answers a purchase question, so blocking grounding there has the highest cost.
  3. Remember that a header applies to the whole response. A rule added “for the blog” at the CDN can silently cover product pages.

AgentReady’s citation-rail check tests whether Bingbot and other citation crawlers can fetch your pages. It does not read response headers, so a store could pass it while a header tells Bing not to ground on the page.

FAQ

Can I block Bing AI training but still appear in Copilot answers?

Yes. Microsoft says training and grounding are independent. You can set AI Training to Disallow in robots.txt and leave AI Grounding on Allow, which keeps the content eligible to inform generated responses without being used to train generative models.

Is there a robots.txt line for Bing AI grounding?

No. Microsoft’s post says AI Grounding is a page-level preference only, set with a meta tag or an X-Robots-Tag header. Only the training preference works in robots.txt.

What happens if I do nothing?

Both preferences default to Allow, so your content stays eligible for training and grounding. The exception is a store that relies on NoArchive or NoCache today: after March 1, 2027 those tags are no longer read as a preference for either use.

Sources

An agent can only buy from a store it can read, quote and trust, so whether a page may be used to ground an answer is now part of the transaction path, not a footnote.