Overview

• Both sides requested a confidential August 7 settlement conference to resolve disputes over the scope of ChatGPT log discovery in the New York Times lawsuit.
• OpenAI is pushing for a 20 million-log sample deemed statistically sufficient, while the news plaintiffs insist on reviewing 120 million individual conversations.
• The company warns that producing and de-identifying 120 million unstructured logs could delay proceedings by months and heighten user privacy risks.
• Judge Ona Wang previously rejected OpenAI’s bid to block all log access, prompting parties to seek a focused settlement on sample size.
• The outcome could set a precedent for balancing large-scale AI data requests with privacy safeguards and technical feasibility.