Meta Platforms, the parent company of Facebook and Instagram, faces a copyright lawsuit alleging the use of thousands of pirated books to train its AI models, specifically the artificial intelligence language model known as Llama. The complaint consolidates two separate lawsuits brought against Meta by notable figures such as comedian Sarah Silverman and Pulitzer Prize winner Michael Chabon, who claim that their works were used without permission.
The lawsuit, filed on Monday, accuses Meta of using pirated books to train its AI models, despite warnings from the company’s lawyers about the potential legal risks. The authors allege that Meta used their works to train Llama without obtaining proper authorization, violating their copyrights.
Evidence presented in the lawsuit includes Discord chat logs of a Meta-affiliated researcher, Tim Dettmers, discussing the procurement of the dataset used to train Llama. In the chat logs, Dettmers describes his communication with Meta’s legal department and expresses uncertainty about the legality of using book files as training data. The logs indicate that Meta was aware of potential legal issues related to copyright infringement.
The term “The Pile” in the chat logs refers to an open-source language modeling dataset that Meta acknowledged using to train its initial version of Llama. The complaint quotes Dettmers mentioning that Meta lawyers told him the data could not be used or models published if trained on that data. The researchers discussed concerns about books with active copyrights, acknowledging that these posed the most significant legal risk.
Llama is Meta’s large language model, and the lawsuit raises questions about the use of copyrighted content to train AI models. The legal action adds to a growing number of lawsuits targeting tech companies for alleged copyright infringement related to AI and machine learning. The outcome of such lawsuits could impact the use of copyrighted material in training datasets and the broader field of generative AI.
Post Your Comments