Publishers Launch New Lawsuit Against Google for Using Digital Books to Train AI
Article excerpt
A new publisher lawsuit about the use of books in AI training is particularly compelling.
A group of three publishers, plus author Scott Turow, has brought perhaps the most straightforward case against a tech company’s use of books in training AI models.
The case is not focused on on the legality of using books at all in training these models, but rather on Google’s specific sourcing of the books it used. Namely, the suit alleges that Google used books in its Google Scholar, Google Books, and the Google Play bookstore outside of the express and agreed upon uses of those platforms. Books were made available in each of those projects for a specific purpose and Google violated copyright when it used those books outside of those parameters.
Furthermore, the suit alleges that Google’s own internal discussions identified using books in these systems opened the company up to litigation.
To my untrained legal eye, this looks to be a strong case with the possibility of significant damages, even as it does not address the question most of are asking: is using books, even if acquired legally, to train AI models fair use.
Publishers Lunch first reported this story. A great read if you don’t subscriber over there already.