Court filing in Authors Guild v. OpenAI/Microsoft. Plaintiffs allege OpenAI downloaded millions of pirated books from Library Genesis (LibGen) for GPT training.
Key claims: Alec Radford downloaded ~117,500 books Oct-Nov 2018; GPT-3 trained on LibGen data April 2019; Microsoft received LibGen datasets under commercial GPT-3 agreement early 2021; OpenAI chose LibGen over licensing despite Microsoft investment.
Plaintiffs seek partial summary judgment on direct infringement. Defendants maintain fair-use defense.