New York Times and Daily News ask judge to sanction OpenAI, alleging it hid evidence in copyright fight

The sanctions motion

A group of news publishers led by The New York Times and the Daily News asked a federal court in Manhattan on Thursday to sanction OpenAI. The motion accuses the ChatGPT maker of discovery misconduct, claiming it lied for two years about its ability to search training data and chat logs for copyrighted material. The publishers say OpenAI "chose obstruction" over transparency, and they want the court to punish the company for hiding and destroying evidence.

For over two years, OpenAI lied to The Times, The Daily News Plaintiffs, the public, and the court.

What the deposition revealed

The allegations rest heavily on an April 2026 court-ordered deposition of OpenAI privacy engineer Vincent Monaco. According to the filing, Monaco testified that OpenAI had already conducted internal searches of its training corpus for copyrighted news content, contradicting the company's earlier claims that such searches were infeasible. He also disclosed that OpenAI had built a database of about 78 million de-identified ChatGPT conversations and implemented a "Bloom" filter, part of a toolset called Project Giraffe, to detect regurgitation of copyrighted material.

It claimed searching ChatGPT outputs for copies of The Times' and the Daily News Plaintiffs' content was infeasible, burdensome, and invasive of users' privacy — while at the same time concealing that it had already done such searches.

Logs deleted and redacted

The publishers also claim OpenAI deleted billions of ChatGPT outputs after the lawsuit was filed, violating a court preservation order. When OpenAI finally produced a sample of 20 million chat logs in December 2025, the sample was so heavily redacted that the court called it "unusable." The plaintiffs say the company substituted millions of logs and made it needlessly difficult to obtain information it had already collected internally.

OpenAI's response

An OpenAI spokesperson called the sanctions motion a late effort to access more logs and invade user privacy. The spokesperson said the Times' case is weakening, pointing to the recent dropping of some claims, and that the company will continue defending fair use and user privacy. The Times' spokesperson, Graham James, countered that the core claims remain unchanged and that the suit was streamlined by adding claims against Microsoft.

As the Times' case weakens and they've been forced to drop claims against us, they're persisting with their efforts to invade the privacy of people who have nothing to do with this case, including by making these blatantly false allegations.

The broader AI copyright battle

The sanctions fight is the latest flashpoint in a case that began in late 2023, when the Times sued OpenAI and Microsoft for using millions of articles without permission to train ChatGPT. The publishers argue that AI chatbots unfairly compete as information sources, siphoning traffic and ad revenue. OpenAI and other tech companies invoke the fair use doctrine, a defense being tested in dozens of lawsuits from authors, artists, and music labels. The Times says it has spent more than $28 million fighting AI companies, while Anthropic agreed to pay book authors $1.5 billion in a separate settlement.