0.13% of the Top 1 Million Websites Use a noai Meta Tag, an AI Opt-Out That OpenAI, Google and Anthropic Don't List in Their Crawler Docs

Summary

Picture an illustrator who reads a forum thread and adds <meta name="robots" content="noai, noimageai"> to her portfolio. She feels safer. But the crawlers she worries about may never look for that tag.

noai and noimageai are robots meta values that some art communities and plugins began using in 2022 as a way to say "do not train AI on this page." They are not part of any standard that the large AI companies have adopted. OpenAI's crawler page (developers.openai.com/api/docs/bots), Google's common crawlers page (developers.google.com/search/docs/crawling-indexing/google-common-crawlers) and Anthropic's crawler help article (support.claude.com, article 8896518) all describe robots.txt as the control. None of the three mentions noai or noimageai (checked October 21, 2026).

We counted how many popular sites use the tag. The answer is small: 1,027 of the top 1 million homepages, or 0.13%. Most of those sites also use robots.txt rules, which suggests many owners already treat the tag as a backup.

What we measured

We used the HTTP Archive October 2026 mobile crawl (httparchive.crawl.pages) for top-1M homepages and read every meta tag recorded in the almanac meta-nodes metric. A site counts when any meta tag's content contains noai or noimageai as a separate word. We also checked X-Robots-Tag response headers and robots.txt rules for eight AI training crawlers.

For the trend we applied the same word match to the meta tags recorded in six earlier crawls between July 2025 and September 2026.

About 1 in 750 top sites uses the tag

Share by popularity rank.

Rates stay between 0.11% and 0.41% across every bucket. The top 1,000 has 3 sites with the tag, which is the highest rate only because the bucket is small.

Most tag users also rely on robots.txt

We checked what else these 1,027 sites do.

Two thirds of these owners also wrote robots.txt rules, the method the major AI companies say they follow. 103 sites send mixed signals by pairing the tag with an llms.txt file that invites AI tools in.

Use has almost doubled since mid-2025, mostly in one framework

Same word match across seven crawls. Platform names show adoption only.

Most of the growth came between January and April 2026. In October, the tag clusters on Nuxt.js sites: 3.1% of Nuxt.js homepages carry it, far above any other platform. That pattern looks like a shared template or module rather than 345 separate decisions.

What this means

If you want to keep your content out of AI training, use robots.txt rules for the specific training crawlers, because that is the method OpenAI, Google and Anthropic document. A noai tag costs nothing to keep, but do not rely on it.

Be careful not to over-block. Training crawlers and the crawlers that fetch pages for AI answers are separate. Blocking the answer crawlers can remove you from AI recommendations altogether. Ask our team for a free AI visibility audit to see whether AI assistants can find and quote you, including the crawler settings that affect SEO for AI Overviews.

Limits

We match noai and noimageai in any meta tag content, mostly name="robots". A small number of matches may come from non-robots meta tags.

Documentation was checked for three companies on October 21, 2026. Other AI crawlers may treat the tag differently, and policies can change.

robots.txt data counts Disallow lines per user agent, not the blocked paths.

Trend counts come from seven crawls; samples differ slightly between months.

Frequently asked questions

Related studies

• AI training opt-out signalsOnly 0.9% of News Homepages Mark Content as Paywalled
• AI training opt-out signalsOnly 0.13% of EU Sites Declare a TDM Reservation for AI
• AI training opt-out signalsOnly 1 in 4,544 Top Sites Sends an AI Opt-Out Header
• AI training opt-out signalsWeb Bot Auth Keys: 0.01% of Top Sites Publish Them