New York Times and Other Publishers Seek Sanctions Against OpenAI Over Evidence Concealment

The New York Times, Daily News, and six other American publishers filed a motion on Thursday in federal court in Manhattan seeking sanctions against OpenAI. The publishers claim the company misled the court for two years about its ability to search its systems for evidence that millions of news articles were used to train its language models.

The case dates back to late 2023, when The New York Times became the first to sue OpenAI and its business partner Microsoft, accusing them of illegally using millions of articles to train ChatGPT without a license or compensation. Other publishers have since joined the suit, making it one of the most significant legal battles over the relationship between generative AI and the media industry.

Evidence concealment claim

The core of the latest motion concerns not the copyright infringement itself, but OpenAI's conduct during discovery. The publishers claim the company had the ability to search its large language models for copyrighted material but assured the court that such a search was technically impossible, while concealing that it had carried out a similar search itself before the first lawsuit was even filed.

This motion asks the court to sanction OpenAI for concealing and destroying evidence showing how ChatGPT was trained on stolen journalism - Steven Lieberman, counsel for Daily News and seven affiliated publishers

The publishers are asking the court to hold OpenAI liable for the legal costs incurred in forcing the disclosure of data they say should have been entered into the record much earlier. OpenAI rejects the accusations and says that restricting access to ChatGPT conversation logs was meant to protect user privacy, not to conceal evidence.

What's at stake

The case carries significance beyond OpenAI's dispute with the publishers. The outcome could set the terms under which AI companies may use journalistic content to train models in the future, and whether the news industry, which has struggled with shrinking revenue for years, will be compensated for the use of its work without consent or payment.

The New York Times has confirmed that it alone has already spent more than $28 million litigating against AI companies, including the case against OpenAI and a separate proceeding against Perplexity. The scale of that spending shows how costly and drawn-out copyright litigation has become in the era of generative AI.

Relevance for the Polish market

For Polish publishers and content creators, the dispute sets an important precedent, as similar questions about compensation for using press material to train models are also emerging in Europe, including a recent ruling by a Munich court against OpenAI over music content. The outcome of the U.S. proceedings could shape the negotiating strategy of European publishers toward AI companies, including the terms of any future content licensing deals.

The court has not yet set a date to rule on the sanctions motion. If the judge sides with the publishers, it could not only saddle OpenAI with costs but also strengthen the publishers' position in the main copyright infringement case, which is still awaiting a ruling.

Sources: TechXplore via Associated Press (techxplore.com), U.S. News (usnews.com).