Shadow Library

A shadow library is an online repository that gathers up huge numbers of books or papers without proper authorization and makes them available for anyone to download. It looks like a library or archive on the surface, but because it never got permission or paid publishers and authors, it's essentially a 'shadow' collection operating outside the law.

This term is getting attention now because court cases have revealed that some companies downloaded massive numbers of books wholesale from shadow libraries to train the AI models that power chatbots. Interestingly, the court focused less on the act of training AI on copyrighted works itself, and more on where those books actually came from.

Think of it like the difference between buying a book legitimately from a bookstore versus getting hold of a copy that someone illegally photocopied and distributed. Studying the content might look the same either way, but whether the book was obtained legally became the central issue in the case.

As one article put it: "The court took issue only with the fact that Anthropic obtained the books by downloading them without authorization from illegal online 'shadow libraries,' while ruling that using copyrighted works as training material is itself lawful." A common misunderstanding here is thinking this ruling declared that AI training on copyrighted books is illegal outright. In fact, it's the opposite. The court didn't object to the training method itself—only to the fact that the books were obtained without authorization from shadow libraries instead of through legitimate channels.