Ex-Meta Researcher Torrented Pirated Books, Deleted It 'Due to Litigation'

A former Meta researcher, Stephen Roller, admitted to torrenting and distributing the pirated Books3 dataset, which contains nearly 200,000 books, during his time at Meta between 2017 and 2023. The dataset was downloaded from the private torrent tracker Bibliotik. Roller deleted his copy of the dataset about a year ago. The admission is relevant to ongoing lawsuits against Meta over the use of Books3 to train its Llama AI models.