Copyright in the Digital Single Market — text and data mining
AI training on copyrighted works is allowed as text and data mining unless the rightholder has reserved the right in a machine-readable way (opt-out). Research organisations have a wider exception.
What it means for AI
Providers of general-purpose AI models must have a copyright policy that respects opt-outs and publish a summary of training content (AI Act, Art. 53). Pure AI output without human creative input is generally not protected by copyright.
What you must do
- Website owners who do not want AI training on their content: state the reservation in a machine-readable way (e.g. robots.txt and terms of use).
- Companies publishing AI content: check that you have rights to inputs (images, logos, texts) and that outputs do not reproduce protected works.
- Do not assume exclusive rights to AI-generated designs, logos or texts — add human creative work and keep records of it.
Examples
It may not be protected by copyright; register it as a trademark and ideally have a designer rework it.
Add a machine-readable reservation (robots.txt for AI crawlers, terms of use). The EU is agreeing common opt-out protocols.
Key dates
Transposition deadline of the DSM Copyright Directive.
General-Purpose AI Code of Practice published, including a copyright chapter.
General information, not legal advice — consult a lawyer for borderline cases. Regulations reviewed on 29 Sep 2026.