OpenAI Faces Sanctions Demand from US News Publishers Over Alleged Evidence Hiding

The New York Times, the New York Daily News and about 15 other U.S. media outlets have filed a motion in a Manhattan federal court seeking sanctions against OpenAI. The publishers allege the AI firm concealed and destroyed evidence about the datasets and ChatGPT conversation logs used to train its models, including deleting billions of outputs and providing a heavily redacted sample of only 20 million logs out of the 120 million originally requested. They claim OpenAI misrepresented its ability to search its training data for copyrighted news content, despite internal tools such as a “Bloom filter” and a project dubbed “Project Giraffe” that tracked the reuse of protected material. The filing says OpenAI “chose obstruction” and violated discovery orders, and the plaintiffs request monetary penalties, attorneys’ fees and adverse findings. OpenAI denies the accusations, citing user‑privacy concerns and the fair‑use doctrine.

The case, first filed by the Times in late 2023, could shape how AI companies train models on copyrighted material and impact the economics of the news industry.