Is it legal to train AI models on copyrighted books? It’s complicated

AI models used in major chatbots are trained on hundreds of millions of books, articles, and online content without authors' knowledge or consent, raising questions about whether this practice is legal. Copyright law hasn't been updated since 1976, so courts must interpret decades-old guidelines when deciding cases that could shape the AI industry's future. A recent court ruling found that training AI on copyrighted works is lawful and comparable to reading, though it penalized one company for obtaining books through illegal online libraries. The legality ultimately hinges on fair use doctrine, which considers whether the use is "transformative" and whether it directly competes with the original work, leaving many questions unresolved as litigation continues.