Oddbean new post about | logout
 A data set of more than 191,000 books has been used without permission to train generative AI systems by Meta, Bloomberg, and others. “Books3” was based on  pirated ebooks, most of them published in the past 20 years. The data set is at the heart of several lawsuits brought against Meta. The Atlantic's Alex Eisner is sharing this database of authors that have been used for training. Story may be paywalled. 

 https://flip.it/MILzXa

#Books #AI #Bookstodon #Meta #Technology