A coalition of American news organisations, including The New York Times and the Daily News, has asked a federal judge in Manhattan to impose sanctions on OpenAI, accusing the company of concealing and destroying evidence related to how its ChatGPT chatbot was trained on copyrighted news content. The motion, filed on Thursday, escalates a legal battle that could determine the boundaries of artificial intelligence and copyright law, with implications that extend far beyond the United States.
The plaintiffs allege that OpenAI, backed by Microsoft, has engaged in what they call “discovery misconduct” by failing to produce datasets and ChatGPT logs that would show how the AI system used millions of news articles. In a court filing, they claim the company “chose obstruction” over transparency, and that a recent deposition of an OpenAI employee contradicted earlier statements about its ability to search for copyrighted material in its training data.
Steven Lieberman, attorney for the Daily News and seven sister publications, said OpenAI had been “making misrepresentations” for two years. “This motion asks the court to punish OpenAI for hiding and destroying evidence showing how ChatGPT was trained on stolen journalism,” he stated.
Fair Use vs. Fair Competition
At the heart of the dispute is whether AI chatbots like ChatGPT are unfairly competing with news outlets by using their content without permission or payment. OpenAI and other tech firms argue that training AI on digitised books, articles, and web content is protected under the “fair use” doctrine of US copyright law. This theory is being tested in dozens of lawsuits from visual artists, novelists, and music labels, with mixed results so far.
The New York Times filed its original lawsuit in late 2023, shortly after ChatGPT’s launch sparked a commercial AI boom. The paper argues that OpenAI and Microsoft are building rival products that profit from its journalism without authorisation. An amended complaint last month sharpened the focus on unfair competition, a distinct legal argument from those brought by book authors in other cases.
In the largest copyright settlement to date, OpenAI rival Anthropic agreed to pay book authors $1.5 billion (€1.35bn) for training its Claude chatbot on their works without permission. That deal underscores the high stakes for the news industry, which has seen advertising revenue erode as readers turn to AI-generated summaries.
The threat became more acute in 2024, when Google introduced AI-generated overviews at the top of search results, further reducing traffic to original sources. The Times has already spent more than $28 million (€25m) on litigation against AI companies, including a separate lawsuit against Perplexity, according to regulatory filings.
Among the sanctions sought are attorney fees to cover the cost of securing what the newspapers call “improperly withheld” evidence. The case is being watched closely in Europe, where regulators are grappling with similar questions. In France, the French regulator has ordered Meta to resume copyright talks with press groups, reflecting a broader push to ensure tech companies compensate news publishers.
The outcome of this US case could influence European policy, as the EU’s Digital Services Act and Copyright Directive already impose obligations on platforms. If US courts reject the fair use defence, it may embolden European publishers to pursue similar claims against AI developers operating in the bloc.
OpenAI has not yet responded to the latest motion. The company has previously maintained that its practices are legal and that it respects copyright. The trial is expected to proceed later this year, barring a settlement.
