Overview
- The New York Times led a coalition of about 16 news organizations that filed a sanctions motion on Thursday accusing OpenAI of misrepresenting its ability to search model training sets and ChatGPT output and of deleting or rendering roughly 20 million conversation logs unsearchable.
- Plaintiffs say a deposition by OpenAI privacy engineer Vinnie Monaco showed the company had built internal search tools, assembled large de‑identified chat datasets and run searches for publishers’ material before the lawsuits began, undercutting OpenAI’s earlier claims that such searches were infeasible.
- The publishers ask the court to punish OpenAI with remedies that include attorneys’ fees, an order barring OpenAI from relying on its company-provided 20 million log sample, adverse inferences that destroyed logs would have favored the plaintiffs, and findings that ChatGPT reproduced their work.
- OpenAI rejects the allegations and says the motion is false, arguing it will defend user privacy and its fair‑use position and disputing claims that it unlawfully withheld or destroyed evidence.
- If the judge grants sanctions the ruling could substantially narrow the trial issues by treating missing logs as proof of copying, increase litigation costs for OpenAI, and influence how AI developers preserve, search and disclose training and output data in future cases.