by Ars Technica, Ashley Belanger — 7 February 2025
A copyright court case raised by book authors against Meta has revealed the company knowingly trained its AI models on pirated books.
Meta has admitted to torrenting a controversial pirated dataset, and now emails have been unsealed in court to reveal the mass scale of the torrent and that the Meta employees involved openly discussed the possible legal risks and ethics of torrenting pirated content on behalf of the company.
Read the full story here