Data study · AI training opt-out signals
Only 0.13% of Top European Websites Declare a TDM Reservation, While 9.8% Restrict AI Training Crawlers in robots.txt
Across 114,166 top-1M homepages on EU country domains, 150 carry tdm-reservation set to 1, the machine-readable opt-out tied to EU copyright rules. 11,139 restrict AI training crawlers in robots.txt instead.
Analysis of 114,166 top-1M homepages on EU ccTLDs and .eu · HTTP Archive, October 2026 crawl
Key findings
- 1
150 of 114,166 EU-domain homepages (0.13%) declare tdm-reservation=1 in a meta tag.
- 2
11,139 of the same sites (9.8%) name an AI training crawler in a robots.txt Disallow rule, about 74 times as many.
- 3
French domains lead with 63 reservations out of 11,117 homepages (0.57%), then Italy with 39 of 12,790 (0.30%).
- 4
Even so, EU domains are more likely than the rest of the web to use the tag: 0.13% versus 0.03% on non-EU domains.
- 5
Only 63 EU homepages point to a TDM policy with a tdm-policy meta tag.
Summary
Picture a small publisher in Lyon who wants her articles kept out of AI training. Her lawyer mentions that EU copyright law lets rights holders reserve text and data mining, as long as the reservation is machine-readable. She asks her web developer how to do that, and he has never heard of it.
The TDM Reservation Protocol, from a W3C community group, gives one answer: a meta tag named tdm-reservation with the value 1, or a tdmrep.json file. It is built for the opt-out in the EU's 2019 copyright directive.
Almost nobody uses it. Of 114,166 popular homepages on EU domains, 150 carry the tag. Far more European sites use robots.txt rules aimed at AI crawlers.
What we measured
We used the HTTP Archive October 2026 mobile crawl (httparchive.crawl.pages) for top-1M homepages and grouped sites by domain ending: the 27 EU country codes plus .eu, against everything else. We read each homepage's meta tags from the almanac meta-nodes metric and counted tdm-reservation=1, tdm-reservation=0 and tdm-policy tags.
For comparison we counted sites whose robots.txt names at least one of eight AI training crawlers in a Disallow rule. HTTP Archive does not fetch /.well-known/tdmrep.json, so file-based reservations are not included.
robots.txt is the opt-out European sites actually use
Both methods, side by side, on the same EU and non-EU sites.
% of homepages
| Group | Homepages | tdm-reservation=1 | % | robots.txt AI training rule | % |
|---|---|---|---|---|---|
| EU domains | 114,166 | 150 | 0.13 | 11,139 | 9.8 |
| Non-EU domains | 658,246 | 210 | 0.03 | 46,992 | 7.1 |
Source: aisearchoptimization.net analysis of 114,166 top-1M homepages on EU ccTLDs and .eu · HTTP Archive, October 2026 crawl
EU sites use the TDM tag at about four times the non-EU rate, so the law may have some pull. But the absolute numbers are tiny next to robots.txt rules.
France and Italy account for most reservations
Countries with at least 4,000 homepages in the sample.
| Domain | Homepages | tdm-reservation=1 | % TDM | robots.txt AI training rule | % robots.txt |
|---|---|---|---|---|---|
| .de | 16,063 | 14 | 0.09 | 1,679 | 10.5 |
| .it | 12,790 | 39 | 0.3 | 1,115 | 8.7 |
| .fr | 11,117 | 63 | 0.57 | 950 | 8.5 |
| .pl | 11,061 | 4 | 0.04 | 1,396 | 12.6 |
| .nl | 6,906 | 7 | 0.1 | 728 | 10.5 |
| .es | 6,589 | 9 | 0.14 | 598 | 9.1 |
| .cz | 5,751 | 0 | 0 | 564 | 9.8 |
| .ro | 4,955 | 1 | 0.02 | 762 | 15.4 |
| .gr | 4,224 | 1 | 0.02 | 320 | 7.6 |
Source: aisearchoptimization.net analysis of 114,166 top-1M homepages on EU ccTLDs and .eu · HTTP Archive, October 2026 crawl
France (0.57%) and Italy (0.30%) are well ahead of Germany (0.09%) and Poland (0.04%). Czech domains, with 5,751 homepages, have none.
Some sites set the tag to allow mining
The tag can also say the opposite. Here is how the values break down across all 772,412 homepages.
| Tag | EU domains | Non-EU domains | Total |
|---|---|---|---|
| tdm-reservation=1 (rights reserved) | 150 | 210 | 360 |
| tdm-reservation=0 (no reservation) | 20 | 28 | 48 |
| tdm-policy (link to licensing terms) | 63 | 86 | 149 |
Source: aisearchoptimization.net analysis of 114,166 top-1M homepages on EU ccTLDs and .eu · HTTP Archive, October 2026 crawl
Most reservations come without a policy link, so a mining tool would learn that rights are reserved but not how to license them.
What this means
If you publish in the EU and want a legal reservation against AI training, a robots.txt rule alone may not be the whole answer. Ask your lawyer whether you also need a machine-readable reservation such as tdm-reservation, and keep your terms of use consistent with it.
If you want AI visibility, check you have not reserved rights or blocked crawlers by accident through a theme or plugin. Not sure what your theme or plugins have switched on? A free AI visibility audit from our team checks those settings and whether AI assistants can quote you.
Limits
HTTP Archive does not fetch /.well-known/tdmrep.json, so file-based reservations are missing; only the meta tag is counted.
Country is inferred from the domain ending. Many European businesses use .com and are counted as non-EU.
robots.txt counts are Disallow lines per user agent, not full-site blocks.
We describe the legal background in general terms; this is not legal advice.
Embed this research
Add the charts from this study to your page. The embed links back here.
<iframe src="https://aisearchoptimization.net/embed/data-studies/tdm-reservation-eu-websites" title="Only 0.13% of EU Sites Declare a TDM Reservation for AI" width="100%" height="720" style="border:0" loading="lazy"></iframe>
Cite this study
APA
Grenyo, K. (2026). Only 0.13% of EU Sites Declare a TDM Reservation for AI. AI Search Optimization. https://aisearchoptimization.net/resources/data-studies/tdm-reservation-eu-websites
Plain link
Only 0.13% of EU Sites Declare a TDM Reservation for AI - AI Search Optimization - https://aisearchoptimization.net/resources/data-studies/tdm-reservation-eu-websites
Frequently asked questions
What is tdm-reservation?
It is a meta tag from the W3C community group's TDM Reservation Protocol. Setting it to 1 states that text and data mining rights are reserved, which is meant to support the opt-out in EU copyright law.
How many European sites use tdm-reservation?
150 of 114,166 top-1M homepages on EU domains (0.13%) in October 2026.
Which countries use it most?
France, with 63 of 11,117 homepages (0.57%), then Italy with 39 of 12,790 (0.30%).
Is robots.txt more common for AI opt-outs in Europe?
Yes. 9.8% of EU-domain homepages restrict an AI training crawler in robots.txt, about 74 times the number using tdm-reservation.
Related studies
- AI training opt-out signalsOnly 0.9% of News Homepages Mark Content as Paywalled
- AI training opt-out signalsnoai Meta Tag Stats: 0.13% of Top Sites Use It
- 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
See where AI search leaves you out
Run the free AI visibility audit. If you want the fixes done for you, the plan is a flat $299/month.