American media organizations claim that OpenAI “concealed and destroyed evidence” of how it trained ChatGPT on copyrighted news content, while legal costs in this landmark copyright battle have already exceeded 28 million dollars.
Media organizations, including The New York Times and Daily News, are asking a federal judge to impose sanctions on OpenAI. This intensifies the legal battle over artificial intelligence and copyright, which could reshape the future of the struggling media industry. The newspapers allege that the developer of ChatGPT is hiding evidence crucial for a potential landmark trial on copyright infringement regarding how OpenAI and its business partner Microsoft created their artificial intelligence systems using millions of news articles.
The central issue is whether AI chatbots unfairly compete as sources of information, diverting traffic from news sites without engaging in the journalistic work associated with news gathering. In a petition filed Thursday with a federal court in Manhattan, it is alleged that OpenAI “chose a tactic of obstruction” instead of providing datasets and system logs of ChatGPT that could demonstrate how the AI system utilized copyrighted news content. The plaintiffs are asking the judge to penalize the company for “procedural violations during the disclosure of evidence” that could have led to the distortion of evidence, noting that recent testimony from an OpenAI employee contradicts the company’s previous statements.
Attorney Steven Lieberman of New York Daily News stated that OpenAI “made false statements” for two years regarding its ability to search for copyrighted content in its training datasets and AI system logs. “This petition asks the court to punish OpenAI for concealing and destroying evidence that shows how ChatGPT was trained on journalistic materials used without permission,” said Lieberman, who represents Daily News and seven of its subsidiary publications.
Legal arguments and precedents
The New York Times sued OpenAI and Microsoft at the end of 2023, about a year after the debut of ChatGPT sparked a commercial boom in artificial intelligence and began changing how information is searched on the internet. The threat to news outlets became even more acute in 2024 when Google introduced AI-generated summaries at the top of search results, leading to a decline in advertising revenue due to reduced user traffic to original sources.
Since then, other media organizations have joined Times, including MediaNews Group (the parent company of Daily News and Chicago Tribune), digital media publisher Ziff Davis, and the nonprofit organization Center for Investigative Reporting. OpenAI and other tech companies argue that training their AI systems on digitized books, online articles, and other web content is protected by the “fair use” doctrine of US copyright law. This theory is being tested in dozens of lawsuits as visual artists, novelists, music labels, and other creative industries sue AI companies with mixed results.
In the largest copyright settlement to date, OpenAI‘s competitor, Anthropic, agreed to pay book authors 1.5 billion dollars (1.35 billion euros) for training its chatbot Claude on their works without permission. The Times‘ arguments differ from those made by book authors. In its original complaint and amended lawsuit filed last month, it focused on the unfair competition of companies seeking to profit from its journalism without permission or payment to create competing products.
Financial consequences and licensing agreements
The New York Times has already spent more than 28 million dollars (25 million euros) fighting AI companies in court. This is evident from regulatory documents disclosing its legal expenses, including a separate lawsuit filed last year against AI company Perplexity. Among the sanctions sought by the plaintiffs is reimbursement for costs, including attorney fees incurred in obtaining evidence that, according to the newspapers, “the company improperly failed to provide.”
The increase in legal costs comes as more media organizations sign licensing agreements with OpenAI and other AI companies, including Google and Meta, which pay publishers for using their news materials or archives to train AI systems.
