The Government Steps Into the Ring
The Trump administration submitted a 20-page legal brief defending OpenAI’s practice of ingesting copyrighted works without authorization to train its large language models. The filing came in response to a lawsuit brought by a major newspaper alleging that the ChatGPT maker infringed on copyrighted books, articles, and other media. In the brief, the administration invoked a presidential executive order on artificial intelligence, arguing that the United States has a strong interest in maintaining a competitive AI industry that sets the global standard.
The document warns that constraining AI development under a narrow reading of copyright doctrine would undermine both scientific progress and economic mobility. By framing the issue as a question of national leadership, the administration elevated what might otherwise be seen as a routine legal dispute into a policy matter with far-reaching implications.
The Fair Use Debate at the Center
Large language models powering chatbots like ChatGPT, Claude, and Gemini are trained on enormous databases of published works. AI companies scrape books, articles, and other media to feed these systems, typically without obtaining permission from copyright holders. Publishers have pushed back, arguing that this practice constitutes infringement rather than legitimate innovation.
The legal crux of the matter is the doctrine of fair use, which allows limited use of copyrighted material without permission under certain conditions. The key question is whether an AI system’s consumption of copyrighted content qualifies as « transformative » — producing something fundamentally new rather than a replica of the original. If a court determines that AI training does not meet this threshold, the financial and logistical consequences for the industry could be enormous.
Recent Court Rulings Favor AI Companies
Judicial outcomes in similar cases have so far leaned in favor of AI developers. Last year, a federal judge ordered Anthropic to pay $1.5 billion in a copyright settlement with a group of authors whose works were used to train the company’s models. However, the penalty was not imposed for the act of training itself. Rather, Anthropic was fined for sourcing those works through illegal shadow libraries — essentially pirated digital collections — not for the AI training process.
The judge drew a striking comparison between how an AI processes text and how a human reader might study a book to learn the craft of writing. « The company’s language models were trained upon works not to replicate or replace them, but to create something different, » the judge wrote, suggesting that transformative intent matters as much as the act of copying.
What This Means for the Future of AI
The administration’s brief is not a judicial ruling, and the executive branch does not have jurisdiction over the case, which remains before the U.S. District Court for the Southern District of New York. Nevertheless, the government’s intervention carries political and legal weight. It signals a clear policy preference for fostering AI development and could sway how judges weigh competing arguments about innovation versus copyright protection.
The outcome of this case will likely set a precedent that shapes how every AI company operating in the United States approaches data sourcing. A broad interpretation of fair use would lower barriers for startups and established firms alike, while a restrictive ruling could force the industry to negotiate costly licensing agreements with publishers and content creators — potentially reshaping the entire economics of artificial intelligence development.
