Meta torrented & seeded 81,7 TB dataset containing copyrighted data

[Costas]

https://arstechnica.com/tech-policy...7tb-of-pirated-books-to-train-ai-authors-say/