Meta Uses 81.7TB of Pirated Works to Train AI



We previously reported that Meta was sued by writers Richard Kadrey, Christopher Golden and comedian Sarah Silverman for allegedly using their intellectual property to train the LLama artificial intelligence (AI) model. According to ArsTechnica, emails sent by Meta employees show that they downloaded at least 81.7TB of pirated works via Torrent. The data downloaded was from the pirated library Library Genesis (LibGen).


In an email that is now evidence in the same case, Meta Research employee Nikolay Bashlykov said that downloading Torrents using company computers was “a perceived wrongdoing.”


Meta previously said that using LibGen data fell under fair use. But prosecutors say that when Meta used Torrent, the data was shared with other pirated users through the seeding feature. Therefore, Meta was involved in the illegal distribution of pirated works.


Library Genesis is a digital library project that contains 2.4 million non-fiction books, 2.2 million fiction books, 80 million science magazine articles, 2 million comics, and 0.4 million magazines. The LibGen site has been blocked and shut down several times for offering free access to pirated works. It was ordered to pay $30 million in damages to publishers last year, although no one knows who the operator is.

Previous Post Next Post

Contact Form