Local News Could Lose in The New York Times vs. OpenAI Case
In a statement of interest filed recently in the case between The New York Times and OpenAI, the Department of Justice urged the court to reject the Times’ argument that training artificial intelligence (AI) models on copyrighted works is itself copyright infringement. If the court allows these broad copyright claims to proceed, there could be negative outcomes for both the AI and local journalism industries.
The Times sued OpenAI in December 2023, alleging that training on its articles without permission infringes on its copyrights. A federal judge allowed the core claims to proceed, and the remedies sought include destruction of the offending models and training data. The government’s position in the statement is that training is a transformative use, distinct from any question about what a model later produces. It argues that creating a broad copyright requirement would hamper competition among AI developers while subsidizing the largest publishers.
But the alleged crisis in American journalism is local, not national. There are now 212 U.S. counties with no locally based news source and another 1,525 with only one, leaving roughly one in seven Americans with limited or no access to local news. The Times counts nearly 13 million subscribers and competes directly with local outlets for the single news subscription most people will ever buy. It is a reminder that the interests of a national incumbent and a county seat weekly diverge, and a legal rule built around the former does not automatically rescue the latter.
The government’s filing now recognizes this dynamic. Licensing fees, it warns, “would disproportionately benefit legacy media outlets due to the sheer volume of their written publications,” and entry barriers built on those fees would “function primarily as large subsidies for old mainstream media companies.” In other words, the case would only direct licensing revenue to the largest media firms, leaving out the small and disappearing local news outlets.
The same logic runs in the other direction as well. The filing argues that AI tools, if not encumbered by licensing barriers, can help level the playing field between mainstream and independent publishers.
Through a $10 million program run by The Lenfest Institute for Journalism, local newsrooms are employing AI fellows to build tools for reporting, audience growth, and revenue. A number of philanthropic initiatives devoted to revitalizing local news have made grants to equip local journalists with AI-powered public records tools and to build newsroom chatbots that answer reader questions from an outlet’s own archive, complete with citations and links back to full articles. These tools lower the cost of covering a school board meeting or filing a records request—exactly what a two-person newsroom needs. A licensing mandate would raise the price and narrow the capability of those AI tools, and the destruction remedy the plaintiffs seek would reach systems that local newsrooms already depend on. Newsrooms that were never party to this case would absorb costs from both directions, losing tools they use today without gaining licensing income tomorrow.
None of this requires courts to decide the case on policy grounds. The fair use question should rise or fall on the law, and publishers of every size remain free to negotiate voluntary access deals however it is resolved—a point the government’s filing itself observes. What the rest of us should resist is the assumption that a victory for the country’s largest newsroom is a victory for all of journalism. The measure of any rule that emerges from this litigation is its effect on the smallest newsroom in America, not the largest. By that measure, the plaintiff’s theory of the case has remarkably little to offer the outlets it claims to protect.