What is TDM opt-out and why does it matter?
When a creative work enters an AI training pipeline, something legally significant happens. The work — a photograph, a piece of music, a written article — becomes data. But the act of calling it data does not extinguish the rights that subsist in it.
Text and data mining opt-out — known as TDM opt-out — is one of the primary legal mechanisms through which rights holders can assert that their content may not be used for AI training without permission.
What the law says
Three major jurisdictions have established distinct frameworks for TDM in the AI context.
United Kingdom
The UK's Data (Use and Access) Act 2025 governs TDM in the UK. Under the Act, research TDM is permitted for non-commercial purposes, but commercial TDM — including AI training — requires either a licence or explicit permission from the rights holder.
European Union
The EU AI Act, particularly Articles 50 and 53, requires AI system providers to document the content used in training and to respect TDM opt-outs applied by rights holders under the EU Copyright Directive.
United States
US law approaches TDM through fair use analysis. While no explicit TDM statute exists at federal level, the Copyright Office's guidance makes clear that training use of copyrighted material is subject to the four-factor fair use test.
How to declare a TDM opt-out
- Add a CIP-TDM-Opt-Out: true declaration to your cip.md file hosted at the root of your domain
- Include a Disallow directive in robots.txt for AI crawlers
- Add machine-readable metadata to your content using C2PA Content Credentials
- Include a TDM Opt-Out Clause in any licensing agreements covering your content
The CIP approach
CIP treats TDM opt-out not as an edge case but as a baseline right. Every creator, agency, and platform operating within the CIP framework is expected to understand TDM opt-out, to declare it where appropriate, and to respect it when encountered on others’ content.