Court Orders OpenAI to Turn Over Internal Communications About Deleting Pirated Book Datasets
Summary
Federal court orders OpenAI to surrender internal communications about deleting pirated book datasets after the AI company waived attorney-client privilege by changing explanations, potentially exposing billions in copyright infringement damages as authors strengthen their case against illegal book piracy used to train AI models.
Key Points
- OpenAI loses a key discovery battle and must turn over internal communications about deleting two datasets of pirated books called 'books 1 and books 2'
- The court rules that OpenAI waived attorney-client privilege by changing its explanations for why the datasets were deleted, opening the door to potentially damaging evidence
- OpenAI faces billions in damages if found guilty of willful copyright infringement, with the discovery ruling strengthening authors' arguments about illegal book piracy