In a class action case filed by the Authors Guild and its co-plaintiff authors against OpenAI and Microsoft, all parties have filed motions for summary judgment. In a summary judgment, a judge may weigh undisputed evidence and render a decision without a full trial being necessary.
In September 2023, the Authors Guild and 17 co-plaintiffs first filed a class action suit against OpenAI and Microsoft. As of October 2025, this case was consolidated with other class action cases, and the book-related components now are being considered alongside a similar case from news organizations, led by the New York Times. The entire docket, now known as In Re: Open AI Inc. Copyright Infringement Litigation (1:25-md-03143), is being heard in the Southern District of New York by U.S. District Court Judge Sidney Stein.
In addition to the Authors Guild, the class action's lead co-plaintiffs presently include authors David Baldacci, Taylor Branch, Michael Connelly, Sylvia Day, Jonathan Franzen, Christopher Golden, Andrew Sean Greer, John Grisham, David Henry Hwang, George R.R. Martin, Jodi Picoult, Stacy Schiff, and James Shapiro. Of these 13 authors, Baldacci, Connelly, Day, Franzen, Grisham, Martin, and Picoult have been parties to the case from the start.
‘Mass piracy’
According to the Authors Guild and co-plaintiffs, OpenAI and its primary shareholder, Microsoft, committed copyright infringement by using fiction writers’ work to train its generative pre-trained transformer, or GPT. Citing copyright law and fair use doctrine, the September 5 request opens with the assertions that “OpenAI’s GPT models pose an existential threat to those who write and publish books” and that this “threatens the incentives to create, the livelihoods of authors, and the future of book publishing.”
The discovery process has high stakes for all parties and has been widely publicized, leading the court to seal many letters and exhibits from public view. The Authors Guild’s filing is heavily redacted, blacking out proprietary information about the number of books affected and the methods by which the class plaintiffs say OpenAI and Microsoft “worked together” to duplicate copyrighted material and assemble training datasets based on authors’ work.
The plaintiffs argue that “OpenAI built the foundations of its business on mass piracy,” which included reproducing and then deleting book content found on the pirate site Library Genesis, aka LibGen. Plaintiffs note that courts have “consistently held that obtaining ‘for free’ something one would ‘ordinarily have to buy’ is not fair use.”
They also maintain that books comprise “high-quality data because they contain organized writing, structured thoughts, and [redacted characteristics]” that enable LLMs to convincingly mimic human storytelling and styles of communication. As to harm being done to the profession, they cite increasing sales of AI-generated fiction on platforms such as Kindle e-readers, and quote OpenAI machine-learning technician Tarun Gogineni’s viral 2025 tweet that, if Song of Ice and Fire series author George R.R. Martin “dies early, GPT-5 will autocomplete his series.”
‘Only 0.62% of’ A Game of Thrones
OpenAI defendants maintain in their September 4 filing that the “alleged use was fair because it was highly transformative” of data fed into its large language models, and “no property right is unlimited.” For the defendants, books were among the data collected because “training a useful LLM requires hundreds of billions to tens of trillions of words from a multitude of diverse sources,” which ostensibly are mined for general linguistic patterns and not to replicate “creative expression.”
They also argue that OpenAI interface ChatGPT “does not display copies” of books and that its “content is no different from commonplace excerpts or summaries that anyone can read for free” online. In addition, they claim that ChatGPT has “an alleged regurgitation rate of 0.00007%,” meaning that it rarely produces word-for-word sequences outside of idioms or public-domain texts such as the Gettysburg Address.
The defendants discuss an OpenAI expert’s search for regurgitated passages from the plaintiffs’ books, saying that the plaintiffs’ “expert spent significant effort trying to brute force OpenAI’s LLMs to reproduce text,” by creating millions of prompts. “The longest span of contiguous text he could generate was a 1,899-word excerpt from A Game of Thrones, which is only 0.62% of the book,” the defendants write.
As far as harm to the author plaintiffs, the defendants assert that future books may be written with or without AI assistance, and “competitive harm from new, non-infringing works is not harm that copyright protects against.” Here, evidence from the plaintiffs’ depositions is largely redacted from the publicly available version of the filing.
Ultimately, the defendants believe their scraping of data amounts to fair use because “Open AI’s LLMs are useful for an enormous range of tasks that have nothing to do with books,” because “OpenAI’s use unlocks substantial public benefits,” and because building “a transformative new technology” is “different from the purpose for which plaintiffs wrote the books.” They suggest their position resembles that in 2015’s Authors Guild v. Google Inc. (Google Books) and 2025’s Kadrey v. Meta Platforms, in which defendants prevailed with fair use arguments.
Despite these claims of transformation and innovation, however, the plaintiffs argue that—unlike in Google Books and Kadrey—the copying of books was “nontransformative.” Not only was secretive copying done “for the purpose of developing a competing commercial product,” they allege, it also meant authors and publishers had no opportunity to earn revenue from AI training licenses. For the plaintiffs, OpenAI’s reproduction of creative work was not justified, even if LLMs require “huge volumes of data,” and the resulting technology threatens the market for “human-authored books.”
Opposition briefs in the case will be due in early October, and reply briefs are due in early November.