Overview
- U.S. Magistrate Judge Ona T. Wang denied OpenAI’s bid to reconsider her November ruling, holding the 20 million‑log sample relevant and proportional.
- The logs will be shared under strict protections, including exhaustive de‑identification and attorneys’‑eyes‑only access, which the court said reasonably mitigate privacy risks.
- OpenAI has nearly finished de‑identifying the dataset and must produce it within seven days of completion, per the court’s directive.
- OpenAI has appealed to District Judge Sidney Stein, arguing privacy and security concerns and claiming most logs are irrelevant, positions the court rejected.
- Publishers, led by the New York Times and other outlets, will analyze the sample for potential reproductions and for issues tied to fair‑use defenses; the sample is less than 0.05% of retained consumer logs.