OpenAI, New York Times Case Poses Key Test for Artificial Intelligence Training Under Copyright Law

OpenAI, Microsoft, The New York Times and a group of prominent authors have laid out sharply differing views on one of the highest-stakes questions in copyright law in their dispute over AI training.

 

The technology companies and copyright owners both asked US District Judge Sidney Stein in Manhattan on Friday to rule in their favour on the companies' "fair use" defence, a decision that is likely to determine which side wins the case.

 

The Times' lawsuit, first filed in 2023, accuses OpenAI and its largest financial backer, Microsoft, of using millions of newspaper articles without permission to train ChatGPT. A group of authors, including John Grisham, Jonathan Franzen and George R.R. Martin, separately sued the companies that same year for using their books to train OpenAI's popular chatbot.

 

Those cases and other related lawsuits were consolidated in New York last year. Dozens of other complaints have also been brought by copyright owners against technology companies over what they describe as the theft of their material to train AI systems.

 

The pending cases will likely turn on whether AI systems make fair use of copyrighted material by using it to create transformative new content. The first two judges to consider the issue issued diverging rulings last year.

 

US District Judge William Alsup in San Francisco called Anthropic's use of books for AI training "quintessentially transformative".

 

Judge Vince Chhabria, also in San Francisco, ruled for Meta in a similar case two days later and also called its use of copyrighted books transformative. Chhabria warned, however, that AI training would not be fair use "in many circumstances" and raised concerns that generative AI could "flood the market" with content that competes with human creators.

 

The authors echoed Chhabria in their filing on Friday, arguing that AI is "diluting the market for books across the board" and that fair use "does not countenance such catastrophic threats to the incentive to create".

 

In their brief, the news outlets said that ChatGPT diverts users from their websites and also displaces the market for their work.

 

"Defendants' competitive exploitation of Plaintiffs' expressive works cannot be excused" under court precedent, the outlets said.

 

Citing Alsup, OpenAI told the Manhattan court that the use of copyrighted works for AI training is "among the most transformative many of us will see in our lifetimes" and said its technology has not harmed writers.

 

"The purpose of OpenAI's pretraining process was to derive broad, unprotectable statistical patterns related to language that can be used to create new text, not reproduce protected expression," OpenAI said in a brief addressing the authors' case.

 

Microsoft also told Stein on Friday that fears that large language models would displace authors and journalists were unfounded.

 

"Plaintiffs began this case speculating that LLMs would destroy their very livelihoods," Microsoft said in a brief addressing the authors' case. "Years of discovery later, the record is clear: Neither LLM training nor the use of LLMs in products substitutes for copyrighted books."

 

The case is In re OpenAI Inc Copyright Infringement Litigation, US District Court for the Southern District of New York, No. 1:25-md-03143.