Can AI legally read your favorite books?

Recent court rulings suggest that simply studying a work to teach an AI is not automatically illegal. In one major case, a judge ruled that an AI company’s method of using books to train its system was legally similar to a student studying literature to build knowledge. However, that same company was penalized with a massive fine specifically because it obtained those books from illegal piracy sites, not because the act of training itself was unlawful. Courts are now trying to distinguish between AI that creates something entirely new and AI that acts as a direct, competing product.

To understand this, look at the concept of fair use, a legal rule that allows people to use copyrighted material without permission for purposes like criticism or education. When an AI company trains a model, it breaks down trillions of words into mathematical patterns rather than saving a library of copies. The core question for judges is whether this is a transformative act—meaning, does the AI do something so different that it deserves protection? Some judges argue that if the AI is a tool being used to create something new, it is similar to a human reading a book. But if an AI is designed to act as a direct, competing product, courts are much more likely to step in and stop it.

We are currently in a period of intense legal uncertainty where one judge’s decision could be challenged or ignored by the next. Because the law does not account for AI, we are watching companies and creators fight it out case by case. This means the rules for the next generation of AI are being set by individual courtroom rulings rather than clear updates to our laws. We are waiting to see if our current copyright system can survive in a world where machines can digest the entire history of human writing.