Glossary
Copyright and AI
Copyright and AI is the open question of who owns what: whether AI companies may train on copyrighted works, and whether AI-generated output can be protected at all. Courts and lawmakers are still settling both, and purely AI-made works generally get no copyright.
Two separate questions hide under this label, and both are still being fought over.
Question one: was it legal to train on copyrighted work? LLMs learned from books, articles and images, many under copyright. Authors, newspapers and artists have sued AI companies in the US and Europe; rulings so far point in different directions, and the EU adds its own layer: a text-and-data-mining exception with an opt-out for rightsholders, plus AI Act transparency duties about training sources. None of it is settled. We describe how things stand; we don’t predict verdicts.
Question two: who owns what the AI makes? The widely shared baseline: purely AI-generated work isn’t copyrightable, because copyright protects human creativity. The US Copyright Office has said so explicitly, and EU law points the same way. Your own creative input can make a work protectable; where exactly the line sits is still being drawn.
Practically: a bar owner can use an AI-made flyer (providers like OpenAI assign you their rights in the output) but she may not be able to stop a rival from making a near-identical one. For logos or anything brand-critical, human authorship still buys real protection. For specific cases, ask an IP professional, not a chatbot.
Where you’ll meet this
- Terms of use of ChatGPT, Claude, Gemini and image tools, in the “ownership of output” sections
- News about lawsuits between publishers, artists and AI companies
- Stock-photo and freelance platforms with rules about AI-generated work