The Trump administration has formally sided with OpenAI in a high-stakes copyright dispute with the New York Times, telling a federal court that AI companies need broad access to copyrighted materials to maintain American leadership in artificial intelligence. The intervention came in the form of a 20-page brief filed with the U.S. District Court for the Southern District of New York, where the Times accuses OpenAI and its partner Microsoft of copying millions of copyright-protected articles to train ChatGPT. The brief does not carry the weight of a judicial ruling, but it signals the executive branch's position in one of the most closely watched intellectual-property cases in recent memory.
“The United States has a strong interest in continuing to develop a robust and competitive artificial intelligence industry that sets the standard for the practice and procedure of AI use globally,” the government wrote. “As such, it is critical for the United States to ‘retain global leadership in artificial intelligence.’” The citation refers to an executive order issued by President Donald Trump last year, underscoring the administration's view that AI policy is a matter of national competitive advantage.
The question at the center of the litigation is whether OpenAI's training of large language models is a fair use of copyrighted material. Fair use is an exception in copyright law that allows limited unlicensed use of protected works for purposes such as commentary, criticism, news reporting, research, and education. Courts typically weigh four factors: the purpose and character of the use, the nature of the original work, the amount used, and the effect on the potential market. In technology, fair-use cases often focus on whether something is “transformative” — that is, whether it adds new meaning or character rather than simply copying.
For AI training, the argument turns on whether absorbing millions of copyrighted books, articles, and other texts into a neural network transforms them enough to be deemed legal. Companies like OpenAI, Anthropic, and Google claim that models learn patterns, style, and factual knowledge from these texts and produce original output that does not replace the original sources. Publishers and authors disagree, arguing that training on protected works without permission is no different from wholesale copying and can undercut licensing markets.
The government brief argues that AI training promotes scientific progress and economic opportunity, warning against a ruling that would constrain the technology's development. “Constraining LLM development under a misunderstanding of fair use doctrine would thwart such creative and scientific progress while hindering American prosperity and economic mobility,” the brief said. It also highlighted the global implications for U.S. AI firms, which compete with state-backed Chinese models and European efforts to impose stricter oversight on high-risk AI systems.
The New York Times lawsuit, filed in 2023, seeks to hold OpenAI responsible for the unauthorized use of its journalism. The newspaper has argued that OpenAI's ChatGPT can recall and paraphrase articles and that the company built a product that competes with news publishers without paying for the underlying content. OpenAI has said it tries to avoid reproducing proprietary articles and points to its partnerships with some publishers that license content.
This legal battle is taking place against an evolving landscape of lawsuits and judgments. In 2025, Judge William Alsup of the U.S. District Court for the Northern District of California ordered Anthropic to pay a $1.5 billion copyright settlement to a group of writers whose books were used to train its models. But the fine was not a clear ruling on whether training on copyrighted books is fair use. Instead, the penalty came from Anthropic's method of acquiring the books through illicit “shadow libraries” that distributed pirated copies online, a practice the court treated as distinct from using lawfully obtained or licensed works for model training.
Judge Alsup's opinion in the Anthropic case contained language that has become central to AI companies' defense. “Like any reader aspiring to be a writer, Anthropic's LLMs trained upon works not to race ahead and replicate or supplant them — but to turn a hard corner and create something different,” he wrote. The analogy of an AI model as a human reader is powerful for some judges and scholars, but it has provoked sharp criticism from content creators, who point out that AI systems are mechanical copies running on a massive scale, not an individual reading a library book.
The government's brief enters this debate on the side of the technology sector, not any particular publisher. It does not resolve the legal question, and its authors have no direct role in deciding the case. But in the absence of clearer statutory guidance, courts have looked to a wide range of evidence and expert testimony when making fair-use determinations. A high-level statement from the executive branch can signal to judges how the national interest weighs in that balance, even if it is not binding.
Copyright has long adapted to new technologies. When photocopiers became popular, publishers feared wide infringement, but courts later recognized fair-use protections that allowed educational and scholarly copying. The recording industry challenged the arrival of audio home taping and, later, digital file-sharing services, with mixed results. AI now presents a new boundary, because models are trained not to reproduce exact copies but to extract statistical patterns from massive collections of work, making it harder to map fair-use norms onto the behavior.
For many legal scholars, the critical question is what the training set does with the underlying expression. If a model can regurgitate passages verbatim, that undermines the fair-use case. If it only learns style and facts, that resembles what human authors themselves harvest from the works they read. The U.S. Copyright Office has begun its own inquiry into the issue, and Congress has held hearings but has not passed comprehensive AI copyright legislation.
The outcome in the New York Times v. OpenAI case could set a precedent that affects not only OpenAI but the broader ecosystem of LLM developers, academic researchers, and startups. A ruling that training on copyrighted works is never fair use would radically constrain the ability of AI companies to improve their systems, possibly forcing them to license content or limit their data to expired works and open-source corpora. A ruling that such training is generally fair use would open the door for AI firms to continue expanding their products but would likely prompt publishers and authors to pursue new legislative remedies.
For now, the government's public position exerts a political and rhetorical influence on the proceedings. It also aligns with a moment in which the White House has proposed more flexible AI rules in an effort to keep the country at the cutting edge. The brief explicitly directs the court to consider innovation and global competitiveness, invoking a policy statement signed by Trump. That frame suggests the administration views the copyright case not simply as a commercial dispute but as a matter of strategic economic policy.
OpenAI and its supporters welcomed the government's intervention, interpreting it as recognition of the legitimate role of fair use in technology development. The New York Times and other critics have said that such arguments put the interests of powerful tech companies ahead of the rights of individual creators. The dispute highlights a growing tension between the generative AI industry and the publishing world, which increasingly sees its professionally reported content as valuable data for training systems that might eventually make publishers less relevant.
The case remains before the U.S. District Court for the Southern District of New York. No trial date has been set, and either side could still ask for a summary judgment, which would allow a judge to decide the fair-use issue without sending the case to a jury. The Trump administration's brief will not dictate the outcome, but it has added a new layer of public attention and political pressure to a case that is already considered a bellwether for how the law treats AI training data.
Source: TechCrunch News