Bing AI grounding blocked: how to fix ai-grounding:disallow

The check is called AI training and grounding preferences. It belongs to the AI crawler access dimension and exists since rules version 0.3.3 (2026-11-06). SeenLens reads three places on the page you scan:

• The <meta name="robots"> tag and the <meta name="bingbot"> tag. The tag name is matched without regard to case.
• The X-Robots-Tag response header. A directive with a crawler prefix counts only when the prefix is bingbot, for example X-Robots-Tag: bingbot: ai-grounding:disallow. A directive with another prefix, such as googlebot:, is ignored.
• The robots.txt group named exactly User-agent: __usage_ai-training. The * group is not used for this token.

A training opt-out alone passes. This covers ai-training:disallow on the page and Disallow in a __usage_ai-training robots.txt group. A training opt-out does not affect AI answers, and the evidence line of the check mentions it.

Microsoft announced two separate AI preferences on 2026-10-30: AI Training and AI Grounding. Grounding means using content as input to help inform AI generated responses in Bing, Copilot and partner integrations. Until the new rules apply, Bing has used noarchive and nocache to limit use of content in its AI answers and training. Many sites set these tags for other reasons, so the check warns and does not fail.

A site that uses noarchive or nocache to keep content out of AI answers must publish the new preferences before 2027-03-01. Bing Webmaster Tools has no temporary training setting.

curl -s https://example.com/your-page | grep -i -o '<meta[^>]*name="\(robots\|bingbot\)"[^>]*>'
curl -sI https://example.com/your-page | grep -i x-robots-tag
curl -s https://example.com/robots.txt | grep -i -A3 "__usage_ai-training"

A line with ai-grounding:disallow explains the fail. A line with noarchive or nocache explains the warning. A Disallow under __usage_ai-training is a training opt-out and does not fail the check.

Remove the directive where it is added. Typical sources are an SEO plugin, a theme, a server header copied from another project, and a CDN rule. Keep ai-training:disallow if you only want to opt out of training.

<!-- Fail: the page opts out of Bing and Copilot AI answers -->
<meta name="bingbot" content="ai-grounding:disallow">

<!-- Fixed: no AI grounding restriction (Allow is the default) -->
<meta name="robots" content="index, follow">

<!-- Training opt-out only: this passes -->
<meta name="bingbot" content="ai-training:disallow">

Remove the same directive from the response header. If the header comes from your server, delete the line that sets it:

# Nginx: delete this line
add_header X-Robots-Tag "Bingbot: ai-grounding:disallow";

# Apache: delete this line
Header set X-Robots-Tag "Bingbot: ai-grounding:disallow"

In Next.js, remove the entry from the headers() function in next.config.js. On Vercel, remove it from the headers array in vercel.json:

// next.config.js: remove this entry
{
  source: "/(.*)",
  headers: [{ key: "X-Robots-Tag", value: "Bingbot: ai-grounding:disallow" }],
}

// vercel.json: remove this entry
{
  "headers": [
    { "source": "/(.*)", "headers": [{ "key": "X-Robots-Tag", "value": "Bingbot: ai-grounding:disallow" }] }
  ]
}

Replace noarchive and nocache only if you want to keep content out of AI answers. In that case publish ai-grounding:disallow or ai-grounding:disallowextended instead, as the next section shows. Google Search Central says Google ignores noarchive, so removing it does not change Google behavior.

1. Run the curl commands again. Neither the meta tags nor the header should show ai-grounding:disallow, noarchive or nocache.
2. Open Bing Webmaster Tools, choose URL Inspection, enter the page URL and run the Live URL test. Confirm that Bing can fetch the live page.
3. Run the SeenLens scan and confirm that the AI training and grounding preferences check passes.
4. Wait for Bing to recrawl the page. Microsoft says it applies the preferences as pages are recrawled.

Scan your page with SeenLens

Keep the directive on a page you deliberately keep out of AI answers, such as a paywalled preview or licensed content. The trade-off is that Copilot will not cite that page in its answers. Use ai-grounding:disallowextended if Microsoft products are acceptable and partner use is not.

• Bing AI grounding and AI training: how to set each
• nosnippet and noindex blocking AI Overviews
• robots.txt blocks AI crawlers
• Add Content Signals to robots.txt
• Bingbot

• Bing Webmaster Tools help: New AI Preferences (AI Training and AI Grounding options, syntax, precedence, timeline; checked 2026-11-06)
• Bing Blog: Your content, your choice, clearer preferences for training and grounding (announcement of 2026-10-30; checked 2026-11-06)
• IETF Internet-Draft draft-kelley-aipref-training-preference-signal (the __usage_ai-training token; checked 2026-11-06)
• Google Search Central Blog: Supporting a shared AI training control for websites (planned support, Google-Extended mapping; checked 2026-11-06)
• Google Search Central: robots meta tag and X-Robots-Tag (Google ignores noarchive; checked 2026-11-06)