Industry & Advocacy News
April 19, 2024
The revelations from the investigation conducted by the New York Times in “How Tech Giants Cut Corners to Harvest Data for AI” show big tech’s flagrant disregard for the importance of books and the talent and hard work of authors. Among the examples of AI companies that violate the rights of authors, the discussion by Meta executives about acquiring Simon & Schuster for $10 per book, solely to obtain book data, stands apart. It not only reveals that tech giants would dismantle a cherished, century-old publishing house for parts without compunction, but also demonstrates their contempt and derision for our literary culture and the importance of books to our knowledge and understanding of one another.
It also shows how little such companies understand about trade or commercial book publishing. The publishers do not own the rights to train AI; the authors do. Buying Simon & Schuster would not have put them in a better legal position. They might have obtained better access to copies so they did not have to rely on pirate sites, but to train legally on Simon & Schuster books, they would have had to attain permission from each author.
The ethical and legal way to acquire books for AI training is to get permission—namely a license—from authors, or in cases where the publisher has acquired the rights, from the publisher. For this reason, we are building a licensing system that will allow authors and publishers to license their books to AI companies if they choose to do so.
Fortunately, big tech’s callous indifference to the law and toward books and culture is not universal. There are many stakeholders across the AI field who consider mass, illegal use of creative works troubling. They know that to build lucrative AI technologies which will in turn displace the very works used as their foundation—and without paying authors and artists a cent—is extremely short-sighted.
For instance, we recently announced our support for Fairly Trained, an organization that certifies AI models that are not trained without permission on copyrighted work. These companies and developers understand the value of human authors and creators, and the need to preserve the markets that enable human expression. It’s a shame that the largest and most profitable companies are the ones rejecting licensing, even though the cost to them would be negligible while the benefits to human creators would be immense.
The Authors Guild sees this as an existential battle for books and authorship. We cannot let big tech’s cynical vision define the value of books, works of art, and human culture. This is as much a fight for accountability as it is for defending quintessential human values.
Statements
Authors Guild Supports Groundbreaking Bill to Create AI Transparency Between Creators and Companies
April 12, 2024
Authors Guild Supports New Fairly Trained Licensing Model to Ensure Consent in Generative AI Training
March 20, 2024
Authors Guild Releases AI Best Practices for Authors
February 29, 2024
