Authors Guild Sues OpenAI Over Training on Copyrighted Books

The Authors Guild, the largest professional writers' organization in the United States, filed a class-action lawsuit against OpenAI today in federal court in Manhattan. The accusation: training the models behind ChatGPT on thousands of copyrighted books without permission or compensation for their authors.

The plaintiffs include some of the biggest names in American publishing: George R. R. Martin, author of the A Song of Ice and Fire series on which Game of Thrones is based, along with John Grisham, Jodi Picoult, David Baldacci, Elin Hilderbrand and other signatories, bringing the total to 17 writers. The suit is filed as a class action, meaning it seeks to represent the interests of a broader group of authors in similar circumstances.

What the lawsuit alleges

The complaint claims OpenAI used the plaintiffs' registered works to train its large language models, including GPT-3.5 and GPT-4, without a license or authorization. The Authors Guild's lawyers describe the practice as copyright infringement on a massive scale, and are seeking both monetary damages and injunctive relief that would bar OpenAI from using that material without a prior agreement.

The underlying technical issue isn't new, nor is it unique to this case: large language models are trained on enormous volumes of text scraped from the internet and other sources, which often include digitized books. OpenAI has never publicly disclosed a full list of the works used to train its models, making it difficult for authors to know for certain whether their texts were part of the training data.

Not the first lawsuit of its kind

The Authors Guild's action joins other suits filed this same summer. In July, comedian Sarah Silverman and writers Christopher Golden and Richard Kadrey had already sued both OpenAI and Meta on similar grounds, alleging their books had been used to train models without consent.

What sets today's lawsuit apart is the institutional weight of the plaintiff bringing it. The Authors Guild has spent more than a century defending the economic interests of writers in the United States, and its direct involvement—alongside bestselling authors like Martin and Grisham—raises the media and legal profile of the litigation within the publishing industry.

Why it matters

The case brings into focus a tension running through the entire generative AI industry: models need to ingest vast quantities of high-quality text to perform well, and much of that high-quality text is copyrighted. Until now, companies like OpenAI have argued that training on copyrighted material falls under the "fair use" doctrine in U.S. law, an exception that allows copyrighted works to be used without permission in certain contexts, such as research or creative transformation.

If the courts ultimately side with the authors, the consequences could be far-reaching—from requiring retroactive licensing payments to changing how AI companies collect and document their training data going forward. If the fair-use argument prevails instead, it would cement the status quo that has allowed OpenAI and its competitors to train on material scraped from the web without negotiating rights case by case.

OpenAI has not yet publicly responded to the specific details of this lawsuit. The litigation adds to a growing legal front in the United States, where visual artists, programmers and now writers are trying to get the courts to define, case by case, where the freedom to train models on public data ends and copyright infringement begins.