Unsealed Briefs in Authors’ Case v. Microsoft/OpenAI

619 points · 606 comments on HN · read original →

Points and comments are a snapshot, not live.

New filings claim OpenAI and Microsoft knowingly used pirated books to train AI.

Unsealed briefs in Authors Guild v. OpenAI show executives knew using LibGen (a "sketchy Russian website") was illegal. OpenAI internal Slack messages discuss deleting LibGen files in "Project Clear" to hide evidence. Policy Director Jack Clark stated in 2020 the work would "make people unemployed." Researcher Tarun Gogineni wanted GPT models to finish George R.R. Martin's series. Microsoft knew of LibGen use as early as April 2019.

What commenters are saying

Most commenters criticize OpenAI's ethics and the disclosure of internal quotes showing they feared HN headlines, not the law. Some defend LibGen as a research library, arguing the article is biased lobbying. Others argue copyright law needs reform but that breaking existing laws is wrong. One commenter laments that AI outputs are now degraded "claudeslop" polluting training data.