What Is the noai Meta Tag?
The noai meta tag is an HTML directive — <meta name="robots" content="noai">, often paired with noimageai — that asks AI systems not to use a page's content for machine-learning training. It originated at DeviantArt in November 2022 during the backlash over art scraped into image-generation datasets, and it remains one of the clearest examples of the gap between declaring a preference and enforcing one.
How is noai supposed to work?
Like all robots meta directives, it rides in the page's <head> (or an X-Robots-Tag HTTP header) and relies on crawlers reading and honoring it:
<meta name="robots" content="noai, noimageai">
The directive's scope is training, not indexing — a page tagged noai can still be crawled for search and appear in results. That distinction matters for GEO: noai is not a visibility kill-switch, it is a licensing statement aimed at dataset builders.
Who actually honors it?
Observance is voluntary, and adoption is thin:
- img2dataset, the open-source tool behind several large image-text datasets, added noai/noimageai compliance after DeviantArt's launch
- Major AI labs (OpenAI, Google, Anthropic) do not document noai support; their published opt-out mechanisms are robots.txt user agents — GPTBot, Google-Extended, ClaudeBot
- Raw scrapers ignore it entirely, which is why serious blocking ends at bot management, not meta tags
This makes noai a statement of intent with legal-signal value — it documents non-consent, which mattered in the AI-copyright litigation wave from 2023 onward — but weak technical force.
What should site owners use instead (or alongside)?
Layer by enforceability: robots.txt rules for documented AI user agents first, since the major labs publicly commit to them; CDN or WAF bot rules for the rest; contractual and licensing terms where content has negotiable value; and noai/noimageai as a low-cost additional declaration. Sites that want AI visibility should do none of the blocking — but should still know the tags exist, because CMS plugins and themes occasionally ship them enabled by default.
Example
A photography portfolio adds noai tags and later finds its images referenced in a dataset audit anyway. The tags didn't prevent collection — but they became evidence of explicit opt-out in the takedown request, which succeeded where the meta tag alone had failed.
Frequently asked questions
- Who introduced the noai meta tag?
- DeviantArt introduced noai and noimageai directives in November 2022, letting artists flag pages as off-limits for AI training. The img2dataset tool — used to build large image-text training sets — added support for respecting them, but observance across the wider crawler ecosystem remains voluntary and patchy.
- Does noai affect my visibility in AI search answers?
- Generally no. noai targets training-data collection, not search retrieval. Answer engines' search crawlers key off robots.txt user-agent rules instead. A page can carry noai and still be retrieved and cited — or lack it and never be.
Keep exploring
See how AI engines talk about your brand — track mentions across ChatGPT, Perplexity, Claude, Gemini and 5 more. Start with Menra