Updates to our Terms of Use

We are updating our Terms of Use. Please carefully review the updated Terms before proceeding to our website.

Tuesday, May 14, 2024 | Back issues
Courthouse News Service Courthouse News Service

Sarah Silverman attorney says he can provide ‘direct evidence’ that ChatGPT infringed her work

Silverman and two other authors claim that ChatGPT was trained using their work without permission.

SAN FRANCISCO (CN) —  Comedian and author Sarah Silverman and two other authors fought Thursday against dismissal of their class action accusing OpenAI of training their ChatGPT program using material from their books without permission, infringing their copyrights and enriching itself in the process.

In a July 2023 complaint against OpenAI, Silverman, along with authors Christopher Golden and Richard Kadrey, claim that ChatGPT can “accurately” summarize their books when prompted, and say they “did not consent to the use of their copyrighted books as training material” for ChatGPT.

The authors seek to represent a class of all U.S persons who own a copyright in any work — not just books — used as training data for the OpenAI Language Models. They say that their work was illegally obtained from “shadow library” websites containing their work such as Bibliotik, Library Genesis and Z-Library. 

OpenAI, looking to dismiss the suit, claimed that the copyrighted work is only a small subset of the data used to train ChatGPT, and that plaintiffs failed to take into account limitations and exceptions, including fair use laws, that “properly leave room for innovations like the large language models now at the forefront of artificial intelligence.”

OpenAI said that the motion should be dismissed because the authors are claiming that every single output of ChatGPT is an infringing “derivative work,” because it is based on copyrighted material.

The Ninth Circuit has rejected that conception of derivative works in the past, according to OpenAI, holding that a derivative work claim requires plaintiffs to show that accused work shares copyright-protected, expressive elements with the original.

In a separate case, Silverman sued Meta over claims that Meta’s LLaMa language model copied her work, but U.S. District Judge Vincent Chhabria dismissed most of it, writing that Silverman didn’t provide evidence that LLaMa’s outputs were substantially similar to her books.

In a motion to dismiss hearing on Thursday, U.S. District Judge Araceli Martinez Olguin asked the plaintiffs’ attorney, Joe Savari, to explain how Chhabria got the Meta decision wrong and why the same standards should not apply to this case.

Savari said that direct enrichment claims still can proceed against OpenAI, and said that substantial similarity is not a requirement under the Ninth Circuit if he can provide “circumstantial or other direct evidence” that OpenAI copied his clients’ material.

Martinez-Olguin replied that the circumstantial or direct evidence was not present in the complaint in front of her. Savari asked for leave to amend the complaint.

“Certainly, we feel with leave to amend, that there are more facts to allege,” Savari said. He said if allowed he would submit a larger document that fixes the current complaint’s shortcomings.

“It’s a promise, not a threat, that it will be a much larger document and we look forward to the opportunity,” Savari said.

Andrew Gass, counsel for OpenAI, said that he agreed with Chhabria’s interpretation of the Meta case and then brought up Universal’s hit movie E.T. the Extra-Terrestrial as an example.

Gass said that in 1984, the Ninth Circuit ruled against playwright Lisa Litchfield, who sued director Steven Spielberg for $750 million, claiming he used her one-act musical play Lokey from Maldemar as the basis for E.T. after Litchfield had tried to sell it to Universal Studios in 1978.

“Everyone agreed that Universal had access to the copyrighted work,” but the Ninth Circuit held that the suit was frivolous, Gass said.

He said that the plaintiffs needed to prove substantial similarity, and that they had failed to meet the burden, thus the Ninth Circuit's standards should hold.

“The plaintiff needs to show that their things look like,” the output from ChatGPT, Gass said.

On rebuttal, Savari again said that he has proof that his clients’ work was copied, and said he thought a trial would be the “most appropriate” way to solve the case.

Martinez-Olguin took the matter under submission.

“I won’t promise you any sort of timeline,” she told the counsel before court adjourned.

Categories / Courts, Technology

Subscribe to Closing Arguments

Sign up for new weekly newsletter Closing Arguments to get the latest about ongoing trials, major litigation and hot cases and rulings in courthouses around the U.S. and the world.

Loading...