On Wednesday, two U.S. authors filed a class action lawsuit against OpenAI in San Francisco federal court, alleging that the business used their writings to “train” ChatGPT.
Paul Tremblay and Mona Awad of Massachusetts claimed ChatGPT violated their copyrights by mining data from hundreds of books.
Author attorney Matthew Butterick declined to comment. OpenAI, a Microsoft-backed private startup, did not immediately respond to a request for comment.
Several lawsuits have challenged AI training material. Source-code owners and visual artists sued OpenAI, Microsoft’s GitHub, Stability AI, Midjourney, and DeviantArt. The defendants claim their systems use copyrighted work fairly.
ChatGPT converses with users’ text instructions. It reached 100 million active users in January, two months after launch, making it the fastest-growing consumer app ever.
ChatGPT and other generative AI systems use massive amounts of internet data to generate content. Books are a “key ingredient” in Tremblay and Awad’s complaint because they provide the “best examples of high-quality longform writing.”
The complaint indicated that OpenAI’s training data included over 300,000 books from unlawful “shadow libraries” that sell copyrighted works.
Awad’s works include “13 Ways of Looking at a Fat Girl” and “Bunny.” In February, M. Night Shyamalan’s “Knock at the Cabin” was adapted from Tremblay’s novel “The Cabin at the End of the World.”
Tremblay and Awad said ChatGPT could provide “very accurate” summaries of their books, showing they were in its database.
A nationwide class of copyright owners whose works OpenAI allegedly abused wants unspecified damages.