Is it legal to train AI models on copyrighted books? It’s complicated
A recent court ruling clarified that training AI models on copyrighted books can be lawful if the material is accessed from illegal sources, as the judge treated it like reading rather than copying, though a $1.5 billion penalty was imposed for piracy. The case highlights the murky intersection of outdated copyright law, fair‑use doctrine, and the massive scale of AI training, leaving authors and companies uncertain about future legal standards.