AI and copyright — who owns generated content?

The problem

Traditional copyright protects creative works by humans. But what if an AI writes a novel, creates a work of art, or composes a piece of music? Who then holds the copyright — the user who gave the prompt, the company behind the AI model, or no one?

Illustration created with Canva AI

What does the law say?

In most countries — including the Netherlands and the US — copyright requires human creative input. An AI cannot hold copyright. This means in principle that fully AI-generated content falls into the public domain — no one owns it exclusively.

In the US, the Copyright Office has repeatedly refused to grant copyright to purely AI-generated works. In Europe, the discussion goes even further.

The training data issue

A second major question: may AI train on protected material? Models like Stable Diffusion and ChatGPT were trained on billions of images and texts from the internet — many of which are copyrighted.

Several lawsuits are currently ongoing:

• Getty Images vs. Stability AI: copyright infringement on millions of photos
• Authors vs. OpenAI: use of books without permission
• Musicians vs. Suno/Udio: use of recordings for music AI

What does this mean for users?

• AI-generated content used for commercial purposes can carry legal risks
• Some AI services (such as Adobe Firefly) only work with licensed data and offer indemnification to customers
• Always add a disclaimer when content is AI-generated
• Check the terms of service of the AI tool for ownership rights to the output

The future

Legislation is lagging behind technology. The EU AI Act touches on copyright tangentially, but specific legislation for AI-generated content is still lacking. Expect clarification in the coming years through lawsuits and new regulations.