Comedian and Writer, Sarah Silverman, Sues OpenAI and Meta for Copyright Infringement

On July 7, 2023, comedian and writer Sarah Silverman, along with writers Richard Kadrey and Christopher Golden, filed copyright infringement suits against OpenAI and Meta for using their copyrighted books to train artificial intelligence (“AI”) programs without their consent. These lawsuits are likely the first of many and may have a significant impact upon a copyright owner’s scope of protection against AI programs.

Defendants OpenAI and Meta are the creators of AI software programs designed to “emit convincingly naturalistic text outputs in response to user prompts.” [1] OpenAI created the AI program “ChatGPT” and Meta—the owner of Facebook, Instagram and Horizon Worlds—created the AI program “LLaMA.” The complaints state that such AI programs are “‘trained’ by copying massive amounts of text [from a training dataset] and extracting expressive information from it.” [2] Accordingly, the AI programs “[rely] on the information it extracted from its training dataset” in order to produce “convincing simulations of natural written language as it appears in the training dataset.” [3]

Plaintiffs Silverman, Kadrey, and Golden filed their complaints in the U.S. District Court Northern District of California, San Francisco Division, claiming that OpenAI and Meta “copied” and “ingested” their copyrighted books as training materials for their respective AI programs “without consent, without credit, and without compensation.” [4]

In the OpenAI lawsuit, the plaintiffs proffer documentary evidence to show that “[when] ChatGPT was prompted to summarize books written by each of the Plaintiffs, it generated very accurate summaries” which is “only possible if ChatGPT was trained on Plaintiffs’ copyrighted works.” [5] The exhibits provide ChatGPT's detailed summaries of Silverman's book The Bedwetter, Golden's book Ararat, and Kadrey's book Sandman Slime.

In the separate Meta lawsuit, the plaintiffs track the sources Meta used to train its LLaMA program in order to prove that their copyrighted books appear in those training datasets. [6] Specifically, the LLaMA training dataset allegedly contains many of the plaintiffs' copyrighted books, which Meta illegally acquired from "shadow library" websites, including Bibliotik, Library Genesis, Z-Library, and Sci-Hub. [7] The complaint includes an exhibit listing such copyrighted books in the LLaMA training dataset.

Both lawsuits contain six counts of violations including, among others, copyright infringement, unfair competition, and negligence. The plaintiffs seek statutory damages, restitution for profits, as well as permanent injunctive relief requiring OpenAI and Meta to refrain from using the plaintiffs’ copyrighted works in their AI programs.

These lawsuits may have a profound impact in the intellectual property world. Specifically, courts may clarify how the defense of fair use [8] applies to AI learning from copyrighted materials. Fair use works as an affirmative defense to copyright infringement by allowing the use of copyrighted works for criticism, comment, news reporting, teaching, scholarship, or research. [9] Courts will have to determine the application of the fair use defense in the context of AI and determine whether a program’s use of copyrighted materials to “learn” information constitutes a defense to infringement.

Ultimately, these lawsuits, and the similar lawsuits likely to follow, against OpenAI, Meta and others, demonstrate the legal risks AI program developers face when using copyrighted works and the scope of legal protections that copyright owners may expect against machine learning technology. As the growth of AI technology continues to rapidly evolve, courts will have to address the legal implications surrounding its use and development, especially the impact on intellectual property rights.

Cullen and Dykman’s Intellectual Property team continues to monitor important developments in trademark and copyright law. Should you have any questions about this legal alert, please feel free to contact Karen Levin (klevin@cullenllp.com) at (516) 296-9110 or Ariel Ronneburger (aronneburger@cullenllp.com) at (516) 296-9182.

This advisory provides a brief overview of the most significant changes in the law and does not constitute legal advice. Nothing herein creates an attorney-client relationship between the sender and recipient.

Thank you to Sharlene Cubelo, a summer associate, who assisted in the preparation of this alert.

Footnotes

[1] OpenAI Compl. ¶ 2; Meta Compl. ¶ 1.

[2] OpenAI Compl. ¶ 2; Meta Compl. ¶ 2.

[3] OpenAI Compl. ¶ 3; Meta Compl. ¶ 3.

[4] OpenAI Compl. ¶ 4, 25; Meta Compl. ¶ 5, 19.

[5] OpenAI Compl. ¶ 5, 42.

[6] Meta Compl. ¶ 19-30.

[7] Id.

[8] 17 U.S.C.A. § 107 (West).

[9] Id.