Meta and OpenAI have been sued by comedian and author Sarah Silverman and novelists Christopher Golden and Richard Kadrey over dual claims of copyright infringement in a US district court.
According to the authors, the companies used their books to train their large language models (LLMs) without authorization.
The companies market their LLMs as powerful AI tools that can automate tasks by replicating human conversation and text.
Delving Into the Lawsuit
Among other things, the lawsuit accuses OpenAI and Meta of training ChatGPT and LLaMA, respectively, on data sets that were illegally acquired. According to Silverman, Kadrey, and Golden, these datasets included their works.
The authors allege that the companies acquired their works from Bibliotik, Library Genesis, Z-Library, and other “shadow library” websites, which are “flagrantly illegal”. The lawsuit also offered a thorough explanation of why the plaintiffs believe that their works were illicitly used in the datasets.
Meta listed the sources of its datasets in a paper detailing LLaMA. One of these happens to be ThePile – put together by a company named EleutherAI. According to the complaint, the EleutherAI paper mentions that ThePile was assembled from “a copy of the contents of the Bibliotik private tracker.”
In the lawsuit against OpenAI, the authors have also produced exhibits that show that ChatGPT summarizes books when prompted, thus infringing on their copyrights.
The first such book exhibited in the claim is Silverman’s Bedwetter, followed by Christopher Golden’s book Ararat and Kadrey’s book Sandman Slim.
Similarly, the other lawsuit against Meta alleges that the trio’s works could be accessed in the datasets used by Meta to train its LLaMA models, a quartet of the open-source AI models introduced by the company in February.
In both lawsuits, the authors have claimed that they “did not consent to the use of their copyrighted books as training material” for Meta and OpenAI’s AI models.
Each of the two lawsuits covers six counts of different types of copyright violations, unjust enrichment, negligence, and unfair competition claims. The plaintiffs seek restitution of profits, statutory damages, and more.
The Lawyers’ Ongoing Battle Against AI Companies
The new lawsuits by the authors are only the latest development in Joseph Saveri and Matthew Butterick’s battle against AI companies. They also represent authors Mona Awad and Paul Tremblay in a similar case against OpenAI. Back in November 2022, the duo teamed up and sued GitHub Copilot – an AI coding assistant that was based on large-scale open-source piracy.
Later, in January 2023, they filed a lawsuit against Stability AI, the company behind AI image generator Stable Diffusion for the unauthorized use of billions of digital images as training material. Around the same time, the company was also sued by Getty Images for a similar reason.
Lawsuits such as these are more than just nuisances for Big Tech companies, which seem to be challenging the limits of copyright.
Share
Krishi Chowdhary
Journalist
