OpenAI Faces Mounting Legal Pressure in Authors’ Copyright Lawsuit
A federal court has advanced key arguments in a copyright infringement lawsuit brought by a group of authors against OpenAI, potentially paving the way for a significant settlement or a landmark trial.
Yesterday, U.S. District Judge Sidney Stein permitted the authors to pursue claims that OpenAI infringed copyright both by training its AI models on copyrighted books and by illegally downloading works from “shadow libraries.” The plaintiffs, which include author George R.R. Martin, initially presented these as connected claims, but the court allowed them to be separated, increasing potential avenues for damages – statutory damages for copyright infringement can reach up to $150,000 per work. This case is part of a broader wave of legal challenges questioning the use of copyrighted material in the development of artificial intelligence.
Notably, the judge also found that responses generated by OpenAI’s ChatGPT could be considered infringing upon the original books. The court cited examples where ChatGPT’s summaries of Martin’s A Song of Ice and Fire series were “substantially similar” to the source material, conveying the “overall tone and feel” of the original work. The decision referenced a ChatGPT-generated outline for a sequel to “A Clash of Kings,” stating a jury could find the output infringes on Martin’s works. Similar arguments led to a $1.5 billion settlement between Anthropic and authors in a related case, demonstrating the potential financial stakes for AI companies; you can learn more about fair use doctrine from the U.S. Copyright Office.
OpenAI argued against the separation of claims, asserting the authors improperly added a new claim to the case without notice. However, Judge Stein disagreed, stating the act of illegal downloading itself constitutes copyright infringement, regardless of whether the works were used for training. The case is expected to proceed to summary judgment, where the court will determine which claims will go to trial, and the outcome could significantly shape the future of AI development and copyright law, as discussed in this EFF report.
The court will now move forward with evaluating the merits of each claim, and a trial date has not yet been set.
It’s been around three years since the first AI copyright lawsuit was filed. The state of play is still unclear, but winners and losers in certain cases are emerging.
So far, one of the losers appears to be OpenAI in a lawsuit from book authors, who have steadily been building a formidable case that may force the tech giant’s hand in forking over a big settlement ahead of trial. Earlier this week, a federal court advanced two new theories of infringement against the Sam Altman-led firm. As it stands, the authors, who include George R.R. Martin, have several outs to winning the case.
In total, the plaintiffs are advancing three different arguments: the first is that the training of AI models on copyrighted books constitutes infringement, a common theory that most creators brought when the first wave of lawsuits were filed; the other relates to a newer argument over the practice of pirating books from shadow libraries that weren’t used for training; and the last is that answers generated by ChatGPT are substantially similar to the books they’re trained on.
The theory over the illegal downloading of books has changed over the course of the litigation. At first, lawyers for the authors directly connected the piracy to OpenAI’s training of its models under a single umbrella. But after the various class actions against OpenAI and Microsoft were consolidated, they separated the theories and alleged that the distinct act of illegally downloading the works, regardless of whether they were used for training, constitutes copyright infringement. OpenAI jumped on the move, arguing that its opposition improperly added an entirely new claim to the case without notice.
Important to note: the one win for authors in another AI copyright case, this one initiated by Andrea Bartz against Anthropic, related to the company illegally downloading millions of books. The decision heavily leaned in favor of Anthropic, but the court greenlit that theory for trial. Anthropic later agreed to pay $1.5 billion to settle the lawsuit, despite the ruling largely siding with the AI company on fair use, which allows creators to build upon copyrighted works without a license. The lawyer heading that case, Justin Nelson, is also the lead lawyer for Martin.
On Tuesday, U.S. District Judge Sidney Stein permitted the separation of the shadow library and training theories.
“The prior class complaints asserted a cause of action for copyright infringement and alleged that OpenAI impermissibly downloaded and reproduced plaintiffs’ books,” the judge wrote. “The fact that many of the allegations in the prior class complaints suggested that the ultimate purpose of the reproduction was to train OpenAI’s LLMs is not dispositive.”
Statutory damages for copyright infringement can reach up to $150,000, but there’s no double-dipping. This means that the authors only need to prevail on one of their theories. With the court letting the shadow library argument to proceed, plaintiffs have another path toward maximizing damages.
And in one of the more surprising findings, Stein also concluded that ChatGPT’s answers can possibly be found to infringe upon the books they’re trained on. It pointed to the chatbot’s summaries of Martin’s A Song of Ice and Fire.
“Members of the Night’s Watch, a sworn brotherhood tasked with defending the realm from threats beyond the Wall (a giant ice structure in the North), are attacked by mysterious and deadly creatures known as the White Walkers, thought to be mere legends,” reads a summary of the book’s setting.
Here, the court said that a reader could “easily conclude that this detailed summary is substantially similar” to the book because it “conveys the overall tone and feel of the original work by parroting the plot, characters and themes of the original.”
Outlines for potential sequels to Martin’s works played a part in the decision. When prompted with an alternative sequel to “A Clash of Kings that diverges from “A Storm of Swords,” ChatGPT answered, in part, “Instead of being betrayed, Robb Stark creates a surprise alliance with Renly Baratheon’s remaining supporters, drastically changing the balance of power in the war.”
The court, which stressed that it’s not opining on fair use, said that a jury could undoubtedly find that the output infringes on Martin’s works.
With the ruling, the scales of the case have tipped in favor of the authors. Summary judgment, where the court decides which claims go to trial, will provide a clearer picture.
