The New York Times, the Daily News and several other news organisations have filed a motion asking a federal judge to sanction OpenAI, accusing the company of concealing evidence about how ChatGPT was trained on copyrighted journalism, in a case that could shape the future of a struggling news industry.
A coalition of news publishers has asked a federal judge in Manhattan to punish OpenAI over what they allege is deliberate obstruction of evidence in a landmark copyright lawsuit concerning how the company built its AI systems using millions of news articles. The New York Times, the Daily News and other outlets filed the sanctions motion Thursday, arguing that OpenAI has withheld datasets and ChatGPT logs that could reveal how the technology used copyrighted material during training. The request follows a recent deposition of an OpenAI employee that publishers say directly contradicts the company’s earlier claims about its ability to search its own training data.
What the Publishers Are Asking For
The filing accuses OpenAI of “discovery misconduct” that could distort the evidence available at trial, and asks the court to impose an adverse inference — a legal mechanism that would allow the judge to presume any missing or withheld evidence would have been unfavourable to OpenAI, should the court find the material was improperly withheld or destroyed. The publishers are also seeking an order requiring OpenAI to cover the legal costs and attorney fees they incurred while trying to obtain the disputed materials. Notably, the sanctions motion is directed solely at OpenAI, even though Microsoft remains a co-defendant in the underlying copyright lawsuit.
New York Daily News attorney Steven Lieberman, who represents the Daily News and seven affiliated papers, said OpenAI has misled the court over an extended period. “OpenAI has been making misrepresentations for two years about its ability to search for copyrighted content in its AI training datasets and logs,” Lieberman said, adding: “This motion asks the court to punish OpenAI for hiding and destroying evidence showing how ChatGPT was trained on stolen journalism.” OpenAI did not immediately respond to a request for comment on Thursday.
The Case So Far
The dispute traces back to late 2023, when the New York Times sued OpenAI and Microsoft roughly a year after ChatGPT’s launch triggered a wider commercial AI boom and began reshaping how people search for information online. Concerns about the threat to news publishers intensified further in 2024, when Google introduced AI-generated summaries at the top of its search results, reducing the advertising revenue publishers typically earn when readers click through to original source articles.
Since the Times filed its original suit, it has been joined by other publishers, including the Daily News and Chicago Tribune parent MediaNews Group, digital media company Ziff Davis, and the nonprofit Center for Investigative Reporting. In its original complaint and an amended version filed last month, the Times has focused its argument on unfair competition, accusing AI companies of seeking to “free-ride on The Times’s massive investment in its journalism by using it to build substitutive products without permission or payment.” That framing differs from the approach taken by authors and other creative professionals in similar disputes elsewhere.
Part of a Wider Legal Battle Over AI Training
OpenAI and other technology firms have defended their practice of training AI systems on digitised books, online articles and other internet writing as protected under the “fair use” doctrine of US copyright law — an argument currently being tested across dozens of lawsuits brought by visual artists, novelists, music labels and other creative industries, with mixed outcomes so far. The largest settlement to date in this area came from OpenAI rival Anthropic, which agreed to pay book authors $1.5 billion over the use of pirated works to train its Claude chatbot, a sum representing a small fraction of Anthropic’s $965 billion market valuation as it prepares for a public listing.
The Times has already spent more than $28 million fighting AI companies in court, according to disclosures filed with financial regulators, a figure that includes a separate lawsuit the paper brought last year against AI company Perplexity. Thursday’s sanctions request specifically seeks attorney fees to cover the cost of pursuing evidence the publishers say was “improperly withheld.”
Impact on the Underlying Case
Should the court grant the sanctions, they would not determine whether OpenAI actually infringed copyright, but could influence how evidence is presented and weighed once the case reaches trial, potentially giving the publishers a procedural advantage. The publishers argue that the alleged discovery failures could otherwise prejudice their ability to demonstrate how copyrighted journalism was used to train ChatGPT and other AI models.
A Parallel Trend of Licensing Deals
The rising legal costs come as an increasing number of media organisations have instead opted to sign licensing agreements with OpenAI and other AI companies, including Google and Facebook parent Meta, typically in exchange for a fee allowing their news feeds or archives to be used for AI training. The Associated Press became the first publisher to strike such a deal with OpenAI in 2023.
