TDM Reservation Protocol Explained: EU's Text and Data Mining Opt-Out Mechanism for AI Training Rights

Quick Summary

What this covers: Complete guide to Text and Data Mining Reservation under EU copyright law including implementation, legal status, and comparison to robots.txt for AI licensing.

Who it's for: publishers and site owners managing AI bot traffic

Key takeaway: Read the first section for the core framework, then use the specific tactics that match your situation.

Text and Data Mining (TDM) Reservation is a legal mechanism under EU copyright law allowing rightsholders to prohibit automated text and data mining—including AI training—on their content. Unlike robots.txt (a voluntary protocol), TDM reservation carries legal weight in the European Union, requiring AI companies to respect opt-outs or face copyright infringement liability. Implemented via machine-readable declarations in robots.txt, meta tags, or Terms of Service, TDM reservation provides EU publishers with stronger protection than US publishers enjoy under voluntary robots.txt conventions. However, adoption remains low, technical implementation varies, and enforcement depends on EU member states' copyright enforcement rigor.

Legal Basis: EU Copyright Directive 2019

Article 4: TDM Exception and Opt-Out

Directive (EU) 2019/790 (Copyright in the Digital Single Market) introduced Article 4, granting publishers the right to reserve TDM rights:

"Rightsholders shall be allowed to adopt measures to ensure the security and integrity of their networks and databases where the TDM activities take place. Such measures shall not go beyond what is necessary to achieve that objective."

Interpretation: Publishers can prohibit TDM (including AI training) via "appropriate measures" such as:

Technical restrictions (robots.txt, server blocks)
Contractual terms (Terms of Service)
Machine-readable notices (TDM reservation meta tags)

What TDM Covers

Text and Data Mining is defined broadly:

Automated extraction and analysis of text
Pattern recognition in datasets
Training AI models on copyrighted content
Web scraping for research or commercial purposes

AI training explicitly falls under TDM because it involves automated content extraction and analysis for pattern learning.

How to Implement TDM Reservation

Method 1: Robots.txt Directive

While not standardized in the original Robots Exclusion Protocol, TDM reservation can be expressed in robots.txt:

User-agent: *
Disallow: /

# TDM Reservation (EU Directive 2019/790)
X-TDM-Reservation: 1

Or targeting specific AI crawlers:

User-agent: GPTBot
Disallow: /
X-TDM-Reservation: 1

User-agent: Claude-Web
Disallow: /
X-TDM-Reservation: 1

Legal effect: In the EU, this signals that the publisher reserves TDM rights, making unauthorized scraping a copyright violation.

Technical effect: AI crawlers honoring TDM reservation should treat X-TDM-Reservation: 1 as an absolute prohibition, stronger than standard Disallow.

Method 2: Meta Tags in HTML

Publishers can embed TDM reservation in HTML <head>:

<meta name="tdm-reservation" content="1">
<meta name="robots" content="notdm">

notdm is a proposed meta tag value indicating "no text and data mining."

Example implementation:

<!DOCTYPE html>
<html>
<head>
  <meta charset="UTF-8">
  <meta name="tdm-reservation" content="1">
  <meta name="robots" content="notdm">
  <title>Article Title</title>
</head>
<body>
  <article>
    <!-- Content -->
  </article>
</body>
</html>

Crawlers parsing HTML should recognize notdm and refrain from mining content.

Method 3: HTTP Headers

Serve TDM reservation via HTTP headers:

X-TDM-Reservation: 1
X-Robots-Tag: notdm

Apache configuration:

<IfModule mod_headers.c>
  Header set X-TDM-Reservation "1"
  Header set X-Robots-Tag "notdm"
</IfModule>

Nginx configuration:

location / {
  add_header X-TDM-Reservation "1";
  add_header X-Robots-Tag "notdm";
}

HTTP headers provide a universal mechanism independent of HTML structure.

Method 4: Terms of Service

Include TDM prohibition in your website's Terms of Service:

Text and Data Mining Restriction

Pursuant to EU Directive 2019/790, Article 4, we reserve all rights to
text and data mining (TDM) on this website.

Automated extraction, scraping, or use of content for training artificial
intelligence models is prohibited without explicit licensing.

Violation constitutes copyright infringement under EU law.

Link to Terms from your footer. Visitors accessing the site arguably accept these terms.

Method 5: Creative Commons License Restrictions

Use Creative Commons licenses with No-Derivatives (ND) clauses:

<footer>
  <p>© 2026 Example Publishing</p>
  <p>Licensed under <a href="https://creativecommons.org/licenses/by-nd/4.0/">CC BY-ND 4.0</a></p>
  <p>AI training prohibited without licensing.</p>
</footer>

CC BY-ND allows sharing but prohibits derivatives—AI training arguably creates derivatives.

Limitation: Legal interpretation of whether AI training violates ND clauses is unsettled.

Legal Enforceability in the EU

Copyright Infringement Claims

TDM reservation transforms scraping into copyright infringement:

Without TDM reservation:

AI company scrapes content
Argues "fair use" or research exception
Legal outcome uncertain

With TDM reservation:

AI company scrapes content despite reservation
Publisher demonstrates clear TDM opt-out
Copyright infringement is clear—damages available

Damages under EU law:

Actual damages + profits gained by infringer
Statutory damages (varies by member state)
Injunctive relief (court order to stop scraping)

Member State Enforcement Variability

EU Directive 2019/790 was implemented differently across member states:

Strong enforcement (Germany, France):

Publishers have successfully enforced TDM reservation
Courts recognize machine-readable opt-outs
Damages awarded for violations

Moderate enforcement (Spain, Italy):

Case law is developing
TDM reservation recognized but few test cases

Weak enforcement (some Eastern European states):

Limited case law
Copyright enforcement generally weaker

Publisher strategy: TDM reservation provides strong protection in core EU markets (Germany, France) even if enforcement elsewhere is weaker.

Limitations and Challenges

Low Adoption Rate

Why TDM reservation is underutilized:

Awareness: Most publishers unaware of the option
Technical complexity: Implementing TDM meta tags requires dev work
Unclear standards: No single accepted implementation (robots.txt vs. meta tags vs. HTTP headers)
Crawler non-compliance: Many crawlers ignore TDM reservation

Estimated adoption: <5% of EU websites implement TDM reservation explicitly.

AI Crawler Non-Compliance

TDM reservation relies on voluntary compliance by crawlers. Non-compliant crawlers scrape anyway, requiring publishers to pursue legal action.

Crawler compliance rates:

OpenAI, Anthropic, Google: Likely honor TDM reservation (no public violations reported)
Smaller AI startups: Compliance uncertain
Unidentified scrapers: Ignore TDM reservation entirely

Enforcement gap: TDM reservation is legally binding but practically difficult to enforce without litigation.

Cross-Border Jurisdiction Issues

TDM reservation protects EU rightsholders under EU law. Non-EU AI companies (e.g., US-based OpenAI) face liability only if:

They operate in the EU (have EU subsidiaries, serve EU customers)
EU courts have jurisdiction
Judgments are enforceable internationally

US-based AI companies may calculate that EU enforcement is unlikely or economically insignificant compared to global AI training needs.

Research Exception Ambiguity

Article 3 of the EU Directive allows TDM for scientific research without rightsholder permission.

Gray area: Does AI research at for-profit companies (OpenAI, Anthropic) qualify as "scientific research"?

Arguments for:

AI model development is research
Results are published in academic papers

Arguments against:

Primary purpose is commercial (selling AI services)
Research exception shouldn't apply to for-profit entities

Legal outcome: Uncertain. Publishers reserving TDM rights should expect AI companies to claim research exceptions.

TDM Reservation vs. Robots.txt

Robots.txt

Legal status: Voluntary protocol with no inherent legal force
Geographic scope: Global (but compliance is voluntary)
Enforcement: Requires arguing broader copyright infringement, trespass to chattels, or CFAA violations
Technical implementation: Universally understood by crawlers

TDM Reservation

Legal status: Legally binding under EU law (Directive 2019/790)
Geographic scope: European Union only
Enforcement: Direct copyright infringement claim in EU jurisdictions
Technical implementation: Less standardized, lower adoption

Combined Strategy

Publishers should implement both:

# Robots.txt (voluntary, global)
User-agent: GPTBot
Disallow: /

# TDM Reservation (legal, EU)
X-TDM-Reservation: 1

This provides:

Global protection via robots.txt (voluntary compliance)
EU legal protection via TDM reservation (enforceable in EU courts)

Licensing Under TDM Reservation

TDM reservation isn't all-or-nothing. Publishers can reserve TDM rights while offering licensing.

Licensing Notice Example

This website reserves all text and data mining rights under EU Directive 2019/790.

For licensing inquiries, contact: [email protected]

Licensing rates:
- Academic research: €500/year
- Commercial AI training: €5,000-50,000/year (depending on scope)

Conditional TDM Permission

Allow specific crawlers via licensing agreements:

User-agent: GPTBot
Disallow: /
X-TDM-Reservation: 1

# OpenAI has a licensing agreement—allowlist via server configuration
# (robots.txt cannot express conditional access based on payment)

Server-side implementation: Check if crawler has valid API key/license, then allow access despite robots.txt block.

Monitoring TDM Compliance

Publishers should audit crawler behavior to detect TDM violations:

Log Analysis

Filter server logs for AI crawler user agents:

grep -E "(GPTBot|Claude-Web|cohere-ai)" /var/log/apache2/access.log

If TDM reservation is implemented but AI crawlers still access content, document violations for legal action.

Honeypot Pages

Create hidden pages not linked anywhere:

https://example.com/tdm-honeypot/

Add to robots.txt:

User-agent: *
Disallow: /tdm-honeypot/
X-TDM-Reservation: 1

Any access to /tdm-honeypot/ indicates a crawler ignoring TDM reservation. Log these violations.

Future of TDM Reservation

Standardization Efforts

The World Wide Web Consortium (W3C) is exploring standardized TDM reservation mechanisms.

Proposed standards:

Unified tdm-reservation meta tag
HTTP header standard
Robots.txt extension

Standardization would increase adoption and crawler compliance.

Expansion Beyond EU

UK: Post-Brexit, UK copyright law mirrors EU Directive 2019/790, preserving TDM reservation rights.

US: No equivalent legal framework. TDM reservation in the US relies on contract law (ToS), not copyright law.

Asia-Pacific: Japan's copyright law includes TDM exceptions, but rightsholder opt-outs are less formalized than in the EU.

Global adoption unlikely unless major AI companies adopt TDM reservation as a voluntary global standard.

Frequently Asked Questions

Does TDM reservation apply outside the EU? No. TDM reservation is legally binding only in EU member states. Outside the EU, it's a signal but not legally enforceable under copyright law.

Can AI companies claim research exceptions to bypass TDM reservation? They can argue it, but case law is developing. For-profit AI training likely doesn't qualify as scientific research under EU law.

Do I need to implement TDM reservation if I already block AI crawlers via robots.txt? Not strictly necessary, but TDM reservation provides additional legal protection in the EU. Implementing both is recommended.

How do I enforce TDM reservation if a crawler violates it? Document violations (server logs, screenshots), send cease-and-desist letter, pursue copyright infringement claims in EU courts.

Does TDM reservation impact search engine indexing? No. TDM reservation targets AI training, not search indexing. Googlebot and other search crawlers are unaffected.

Can I reserve TDM rights for some content but not others? Yes. Implement TDM reservation on specific directories:

User-agent: *
Disallow: /premium/
X-TDM-Reservation: 1

Is TDM reservation recognized by all AI companies? No. Compliance varies. Major AI labs (OpenAI, Anthropic, Google) likely honor it; smaller companies may not.

TDM reservation provides EU publishers with legally enforceable protection against AI training—stronger than robots.txt alone. Combined with robots.txt, server-level blocking, and licensing strategies, TDM reservation forms part of a comprehensive AI crawler control framework for EU-based or EU-serving publishers.

When Blocking AI Crawlers Isn't the Move

Skip this if:

Your site has less than 1,000 monthly organic visits. AI crawlers aren't your problem — getting indexed by traditional search is. Focus on content quality and link acquisition before worrying about bot management.
You're running a personal blog or portfolio site. AI citation of your content is free exposure at this scale. Blocking crawlers costs you visibility without protecting meaningful revenue.
Your revenue comes entirely from direct sales, not content. If your content isn't the product (e-commerce, SaaS with no content moat), AI crawlers are neutral. Your competitive advantage lives in the product, not the pages.