The New York Times, the Daily News, and other media organizations are urging a federal judge to impose sanctions on OpenAI, intensifying a legal confrontation over artificial intelligence and copyright. This dispute could significantly impact the future of the news industry.
The plaintiffs allege that OpenAI, the creator of ChatGPT, is withholding crucial evidence. This evidence is pertinent to a potentially landmark copyright infringement trial concerning how OpenAI and its partner Microsoft developed their AI technologies using millions of news articles.
The core of the legal argument centers on whether AI chatbots function as unfair information competitors by diverting web traffic without undertaking the journalistic work of news gathering.
In a filing submitted Thursday at a Manhattan federal courthouse, the news outlets accused OpenAI of choosing "obstruction" over releasing datasets and ChatGPT logs. They contend these materials could reveal how the AI system utilized copyrighted news content. The plaintiffs are seeking penalties against OpenAI for "discovery misconduct" that they claim could distort evidence. They also noted that a recent deposition of an OpenAI employee appears to contradict the company's prior statements.
Steven Lieberman, an attorney representing the Daily News and its sister papers, stated that OpenAI has made "misrepresentations" for the past two years regarding its capacity to search for copyrighted content within its AI training datasets and logs. Lieberman asserted that the current motion seeks to penalize OpenAI "for hiding and destroying evidence showing how ChatGPT was trained on stolen journalism."
OpenAI did not immediately respond to a request for comment.
The New York Times initiated its lawsuit against OpenAI and Microsoft in late 2023, approximately a year after ChatGPT's debut triggered a commercial AI boom and began altering how individuals search for information online. The threat to news publications became more pronounced in 2024 when Google introduced AI-generated summaries at the top of its search results, diminishing the advertising revenue generated when users click links to the original sources.
Since then, The New York Times has been joined by other news organizations in its legal action. These include MediaNews Group-owned newspapers such as the Daily News and the Chicago Tribune, the digital media publisher Ziff Davis, and the nonprofit Center for Investigative Reporting.
OpenAI and other technology companies have maintained that the process of training their AI systems on digitized books, online articles, and other text found on the internet is protected under the "fair use" doctrine of U.S. copyright law. This legal principle is currently being tested in numerous lawsuits involving visual artists, novelists, music record labels, and other creative industries suing AI companies, with varying outcomes.
In a separate case, OpenAI's rival Anthropic reached a settlement with book authors for $1.5 billion to resolve claims regarding the use of their works in training its chatbot Claude. This settlement amount is a small fraction of Anthropic's market valuation as it prepares for a public offering.
The New York Times' legal arguments diverge from those presented by book authors. In its initial complaint and an amended version filed last month, The Times emphasized the issue of unfair competition, arguing that companies are attempting to "free-ride on The Times’s massive investment in its journalism by using it to build substitutive products without permission or payment."
According to filings with financial regulators, The New York Times has incurred over $28 million in legal costs fighting AI companies. These expenses include its ongoing litigation against AI company Perplexity. Among the sanctions the newspapers are requesting on Thursday are attorney fees to cover the costs associated with securing "improperly withheld" evidence.
These increasing legal costs come at a time when a growing number of media organizations are entering into licensing agreements with AI companies like OpenAI, Google, and Meta. These deals typically involve a fee paid to the news outlet for the right to train AI systems on their news archives or feeds. The Associated Press was the first to announce such a partnership with OpenAI in 2023.