Sarah Silverman Sues OpenAI and Meta for Copyright Infringement

2023-07-10

Comedian and writer Sarah Silverman, along with writers Christopher Golden and Richard Kadrey, have filed a dual copyright infringement lawsuit against OpenAI and Meta in a US district court. The lawsuit alleges that OpenAI's ChatGPT and Meta's LLaMA chatbots were trained on illegally obtained datasets that contain the plaintiffs' works. These works are said to have been illegally obtained from "shadow library" websites such as "Bibliotik," "Library Genesis," and "Z-Library," with the plaintiffs pointing out that these books can be obtained in bulk through a seed system. Christopher Golden and Richard Kadrey have both declined to comment on the lawsuit, while Sarah Silverman's team has not responded as of the time of this news release. In the lawsuit against OpenAI, the plaintiffs provide several pieces of evidence showing that when asked to summarize their books, the chatbots infringe upon their copyrights. The first book summarized in the demonstration is Sarah Silverman's "Bedwetter," followed by Christopher Golden's "Ararat" and Richard Kadrey's "Sandman Slim." The plaintiffs claim that the chatbots fail to "replicate any copyright management information contained in the plaintiffs' works." As for the separate lawsuit against Meta, they claim that the writers' books are accessible in the training datasets used by Meta for their four open-source AI models, LLaMA, which were launched in February. The plaintiffs detail in the complaint why they believe these datasets have illegal sources. In the description of Meta's LLaMA, the company mentions one of the sources of their training dataset called ThePile, which is a combination of content from a company called EleutherAI. The complaint states that EleutherAI's paper claims that ThePile was copied from the content of the "Bibliotik private tracker." The complaint mentions that Bibliotik and other listed "shadow libraries" are "clearly illegal." In both lawsuits, the plaintiffs claim that they "did not consent to their copyrighted books being used as training material for the companies' AI models." Their lawsuits include six different claims of copyright infringement, negligence, unjust enrichment, and unfair competition. The plaintiffs seek statutory damages, disgorgement of profits, and more. Lawyers Joseph Saveri and Matthew Butterick, representing the three writers, stated on their LLMlitigation website that they have heard the voices of "writers and publishers concerned about the ability of ChatGPT to astonishingly generate text that resembles copyrighted material, including thousands of books." Saveri also represents programmers and artists in lawsuits against AI companies. Getty Images has also filed an AI lawsuit, accusing Stability AI of developing the AI image generation tool Stable Diffusion, which was trained on "millions of copyrighted images." Saveri and Butterick also represent writers Mona Awad and Paul Tremblay in a similar lawsuit involving the company's chatbot. These types of lawsuits are not only a headache for OpenAI and other AI companies, but they also challenge the limits of copyright. As we have mentioned multiple times on The Vergecast, whenever someone starts talking about copyright law, we see lawsuits centered around these issues that can last for years. Source: https://www.theverge.com/2023/7/9/23788741/sarah-silverman-openai-meta-chatgpt-llama-copyright-infringement-chatbots-artificial-intelligence-ai

RECENT AI NEWS

RECENT AI TOOLS