noai / noimageai directives - the community-convention AI opt-out tokens
Tests for noai, noimageai, or noml tokens in meta robots tags or X-Robots-Tag headers.
What this signal tests
We check whether your pages declare AI opt-out using community-convention robots tokens: `noai`, `noimageai`, and `noml`. These appear either in a `<meta name="robots">` tag or in an `X-Robots-Tag` HTTP response header, alongside the more familiar `noindex`/`nofollow`. The convention originated at DeviantArt in 2022 and spread through artist communities and stock-image sites.
Why it matters for your visibility in AI
Honesty first: this is the least-mature standard in our catalogue. It is a community convention, not an IETF or W3C standard, and it is not honoured by every major AI vendor. But it has become a recognised cultural signal - a way for sites to declare a stance even if enforcement is uneven. For creator-facing platforms, artist portfolios, and sites where the user-facing message matters, publishing noai signals to your community that you are on their side, regardless of which crawlers comply. We report this as a posture signal, not a pass/fail with strong consequences. The practical takeaway: if your site's declared training stance includes opting out, noai should be present alongside the more enforceable mechanisms (TDMRep, aipref, vendor-specific robots rules). If you are happy to allow training, leaving noai off is correct - you should not add it just for the sake of having every checkbox ticked.
Pass criteria at a glance
| Criterion | Passes when |
|---|---|
| Reported as posture, not pass/fail (full pass when site's declared training stance matches noai presence). |
How we test it
We parse the HTML head of a sample of your pages looking for `<meta name="robots" content="...">` tags, and we read the `X-Robots-Tag` HTTP response header on the same pages. We tokenise the contents and check for any of `noai`, `noimageai`, or `noml`. The signal is reported as "declared" or "not declared" - and a pass means the declared stance matches what you have told us (or what the site context implies). We do not flag absence as a failure if your stance is to allow training.
Show technical detection method
Parse meta robots and X-Robots-Tag for noai/noimageai/noml.
If your site fails: how to fix it
- Decide your stance. If you are happy to allow AI training, skip this signal entirely - noai exists for sites that want to opt out.
- If opting out aligns with your policy, add `<meta name="robots" content="noai, noimageai">` to the head of every page. For image-heavy sites add `noimageai` even if you allow `noai`.
- Mirror the same directive in HTTP headers with `X-Robots-Tag: noai, noimageai` from your origin or CDN, for crawlers that skip HTML parsing.
- Do not rely on noai alone. Add the more enforceable signals - TDMRep (see related signal), aipref Content-Usage, and per-vendor robots rules - alongside it.
- Document your stance in a human-readable policy page so users and journalists who notice the directive can find the rationale. The convention's history is at https://www.deviantart.com/team/journal/A-New-Directive-for-Content-Aggregators-934500371.
Quick facts
| Maturity | EMERGING |
|---|---|
| Weight | low |
| Category | Emerging Standards |
Primary sources
Related signals
Frequently asked questions
Is noai a real standard?
No - it is a community convention, originally proposed by DeviantArt in 2022 and adopted by parts of the creator community. It is not an IETF or W3C standard and there is no central registry.
Which AI systems honour noai today?
Adoption is patchy. Some image-training crawlers honour `noimageai`; some text-training crawlers honour `noai`. Major commercial AI labs vary. Treat it as one signal among many, not a guaranteed block.
Why include this at all if compliance is inconsistent?
Because the cost of adding it is near zero, the cultural signal to your users matters, and it costs nothing to layer alongside enforceable mechanisms like TDMRep. Defence in depth.
Is there a stronger alternative?
Yes - TDMRep (legal weight in the EU) and the IETF aipref draft (cross-vendor future standard) are both stronger. Use noai as a third layer, not as your primary mechanism.
Run your own scan
Run a free scan and see how your site grades across all 155 AI-readiness signals.