OpenAI's ChatGPT Was Built on Concealed 'Mass Piracy', Authors Tell Court * TorrentFreak

Key Points:

• Authors have filed a motion for summary judgment in a New York case against OpenAI and Microsoft, alleging unauthorized copying of 194 book titles without permission and arguing this use cannot be considered fair use.
• The authors accuse OpenAI of training its GPT models using pirated books sourced from the illegal Library Genesis (LibGen) torrent site, which OpenAI allegedly tried to conceal by renaming the datasets and later deleting the files due to legal concerns.
• The motion highlights that OpenAI aimed to replace human authors, citing tweets from an OpenAI employee who expressed intent to have GPT models complete works by author George R.R. Martin, illustrating the potential displacement of writers.
• While courts have partially recognized model training as fair use, the authors emphasize that using pirated books is "inherently infringing," citing a 2025 ruling that distinguished lawful training data from illegally obtained content.
• The lawsuit also targets Microsoft, claiming vicarious liability due to its significant investments in OpenAI and potential supervisory role, while OpenAI counters with its own motion asserting fair use and minimal direct copying in its models.