General Purpose AI Models can’t ignore copyright in the EU

Artificial intelligence has posed serious challenges to our understanding of copyright. Broadly speaking, general-purpose models are trained on massive amounts of text and images in order to produce handy tools that can generate answers to a vast array of prompts.

That data could very well be other people’s work. When you ask ChatGPT a question, it is putting to use this “education” to come up with a result. But the result isn’t an exact copy of the curriculum, only a reflection of it. What does that mean for the artists, writers and academics whose work has been put into the curriculum? Should they receive some sort of compensation for their work?