According to TechSpot , a group of authors in the US has filed a lawsuit against Meta, accusing the company of illegally using books to train generative AI models. While Meta denies copyright infringement, internal emails that have just been made public show that some of the company's senior staff and engineers discussed downloading pirated books to serve the AI training process.
Meta AI is embroiled in a legal dispute over internal evidence of downloading and distributing pirated books to train its AI.
The lawsuit documents show that Meta used controversial datasets, including “LibGen,” an archive of millions of pirated books. Meta had previously argued that its use of the dataset fell under “fair use.” However, new emails reveal that Meta not only downloaded but also distributed the datasets via the BitTorrent network, raising serious legal issues.
According to the emails, Meta downloaded and shared at least 81.7 terabytes of data from multiple pirated book repositories, including 35.7 terabytes from Z-Library and LibGen. The plaintiffs called it an “incredible torrent campaign” in which Meta not only downloaded but actively distributed these data files on a large scale.
An internal email from April 2023 from Meta researcher Nikolay Bashlykov noted: “Torrents on company computers seem to be not working.” Although the statement ended with a laughing emoji, his tone changed markedly just a few months later. In September 2023, Bashlykov said he contacted Meta’s legal department because his use of torrents—which is essentially “seeding” pirated data—is clearly against the law.
The documents also show that Meta CEO Mark Zuckerberg was aware of the company's use of LibGen. To avoid detection, Meta reportedly deployed servers outside of Facebook's main system to hide its torrenting and data distribution activities. Another internal email from employee Frank Zhang refers to this tactic as "stealth mode."
Meta is currently investing heavily in the development of AI and services related to generative AI. The company is also looking to integrate chatbots and AI characters into its social media ecosystem. However, with the newly revealed evidence, Meta may have a difficult time defending its “fair use” argument in the face of lawsuits from authors.
Source: https://thanhnien.vn/meta-bi-cao-buoc-dung-noi-dung-sach-vi-pham-ban-quyen-de-huan-luyen-ai-185250209012630928.htm
Comment (0)