NYC Judge Orders OpenAI to Submit Communications with Lawyers Regarding Deleted Databases
A recent ruling by a federal judge mandates that OpenAI must disclose its internal communications regarding the deletion of extensive collections of pirated books from a controversial online “shadow library” that allegedly informed the training of its AI model, ChatGPT. This decision was made by Magistrate Judge Ona Wang in the Manhattan Federal Court as part of a larger class-action lawsuit that implicates OpenAI and Microsoft in copyright infringement activities.
Judge Wang’s ruling, issued on a Monday, determined that OpenAI’s evolving explanations for the removal of the datasets rendered claims of attorney-client privilege ineffective. The judge stressed the importance of transparency, indicating that a jury requires insight into the rationale behind OpenAI’s claim of acting in good faith concerning alleged copyright violations. This legal action has garnered significant attention from numerous media organizations, including the Daily News and various affiliated newspapers, which are collectively pursuing claims against the tech giants.
The ongoing lawsuit includes high-profile plaintiffs such as the Authors Guild and notable authors, including George R.R. Martin, known for the “Game of Thrones” series, and John Grisham, a prolific legal thriller writer. These authors contend that OpenAI improperly utilized materials sourced from the notorious LibGen online library, which has been the subject of legal actions ordered to cease its operations. Allegations include that an OpenAI employee downloaded these pirated books in 2018, ultimately leading to the controversy surrounding the training of the company’s AI products.
During the discovery phase, the plaintiffs uncovered that OpenAI had erased two significant datasets known as “Books1” and “Books2” in 2022, an act that occurred prior to any litigation becoming a factor. Reports suggest that these datasets collectively encompassed over 100,000 books. Initially, OpenAI justified the deletions by citing reasons of “non-use.” However, subsequent inquiries into these deletions led the company to claim attorney-client privilege—a position that Judge Wang indicated has fluctuated over time.
As part of her ruling, Judge Wang has instructed OpenAI to provide the plaintiffs with previously withheld communications related to the dataset deletions, as well as any internal references about the LibGen library that were deemed privileged. While OpenAI’s spokesperson expressed disagreement with the ruling and indicated an intention to appeal, the case continues to unfold, highlighting critical aspects of copyright law in the era of artificial intelligence.
This legal battle raises important questions about the ethical use of copyrighted materials in training AI systems and the obligations of tech companies to comply with copyright laws. The outcome of this case could have far-reaching implications for the future of artificial intelligence development and its intersection with intellectual property rights.
