Judge slices Sarah Silverman copyright suit against OpenAI

SAN FRANCISCO (CN) — A federal judge tossed vicarious copyright infringement claims late Monday night from comedian and author Sarah Silverman and two other authors who claimed OpenAI trained its ChatGPT program using material from their books without permission.

In a July 2023 complaint against OpenAI, Silverman and authors Christopher Golden and Richard Kadrey claimed ChatGPT can accurately summarize their books when prompted, and say they didn't give OpenAI permission to use their copyrighted books to train ChatGPT. 

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

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

At a motion to dismiss hearing in December, OpenAI argued the copyrighted work is only a small subset of the data used to train ChatGPT, and that the plaintiffs failed to take into account limitations and exceptions including fair use laws that allow the use of the material. 

OpenAI sought dismissal of all causes of action except for the claim for direct copyright infringement. The company argued the authors claim every single output of ChatGPT is an infringing “derivative work” but have failed to show that the accused work shares copyright-protected, expressive elements with the original works.

U.S. District Judge Araceli Martinez-Olguin ruled late Monday that vicarious copyright infringement claims fail because the protection does not extend to every idea, theory and fact underlying a copyrighted work, and that the plaintiffs had failed to prove ChatGPT directly copied their work.

“Because they fail to allege direct copying, they must show a substantial similarity between the outputs and the copyrighted materials. Plaintiffs’ allegation that ‘every output of the OpenAI Language Models is an infringing derivative work’ is insufficient,” Martinez-Olguin wrote. “Plaintiffs fail to explain what the outputs entail or allege that any particular output is substantially similar — or similar at all — to their books. Accordingly, the court dismisses the vicarious copyright infringement claim with leave to amend.”

The plaintiffs also claimed OpenAI knowingly removed copyright management information such as the title, author, copyright owner, the terms and conditions for use of the work, and other identifying information when it trained ChatGPT, violating the law.

Martinez-Olguin tossed those claims because the plaintiffs didn't provide evidence OpenAI knowingly removed any copyright management information for the purpose of inducing, enabling, facilitating or concealing infringement.

“Indeed, the complaints include excerpts of ChatGPT outputs that include multiple references to plaintiffs’ names, suggesting that OpenAI did not remove all references to ‘the name of the author,’” Martinez-Olguin wrote.

"Even if plaintiffs provided facts showing defendants’ knowing removal of CMI from the books during the training process, plaintiffs have not shown how omitting CMI in the copies used in the training set gave defendants reasonable grounds to know that ChatGPT’s output would induce, enable, facilitate, or conceal infringement," she continued.

The authors also claimed OpenAI’s failure to state which internet books it uses to train ChatGPT shows that it knowingly enabled infringement, because ChatGPT users will not know if any output is infringing. Martinez-Olguin rejected that for failure to state a claim.

“Plaintiffs do not point to any case law to suggest that failure to reveal such information has any bearing on whether the alleged removal of CMI in an internal database will knowingly enable infringement,” Martinez-Olguin wrote

The authors also brought unfair competition law claims against OpenAI, saying the company engaged in unlawful business practices and violated the Digital Millennium Copyright Act. But Martinez-Olguin tossed those claims because the plaintiffs had failed to show that OpenAI reproduced and distributed copies of their books.

“Accordingly, any injury is speculative, and the unlawful prong of the UCL claim fails for this additional reason,” Martinez-Olguin wrote.

The fraudulent claim under the unfair competition law suffered a similar fate, but the judge advanced the unfair claim because California courts have interpreted unfairness broadly and the plaintiffs made a sufficient showing of unfairness at this stage in the litigation.

Martinez-Olguin gave the authors until March 13 to file an amended complaint.

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.

Email *

Submit