The federal government has formally intervened in the high-stakes legal confrontation between the New York Times and OpenAI, signaling a decisive shift in how the United States intends to balance the protection of intellectual property with the rapid advancement of artificial intelligence. In a significant move on Tuesday, the Department of Justice, representing the Trump administration, submitted a letter to the U.S. District Court for the Southern District of New York, arguing that the training of Large Language Models (LLMs) on copyrighted materials generally constitutes "fair use" under existing law. The government’s position posits that constraining the development of AI tools through overly restrictive copyright interpretations would not only stifle scientific progress but also jeopardize the nation’s standing as a global leader in the burgeoning technology sector.
The Core of the Federal Intervention
The Department of Justice’s letter to the court emphasizes that the United States maintains a "strong interest" in the outcome of this litigation. According to federal lawyers, the legal definition of fair use is central to whether the domestic AI industry can "retain global leadership in artificial intelligence." The administration’s argument rests on the premise that the process of training an AI—ingesting vast quantities of data to learn the statistical relationships between words and concepts—is fundamentally "transformative."
In legal terms, a "transformative" use is one that adds something new, with a further purpose or different character, and does not substitute for the original use of the work. The government contends that OpenAI’s tools do not merely mirror the New York Times’ articles but instead use them as raw material to build a sophisticated system capable of generating entirely new insights and responses. By categorizing AI training as transformative, the administration is aligning itself with the tech industry’s view that LLMs are more akin to a human student learning from a library than a photocopier reproducing a book.
The government’s brief further asserts that the New York Times’ interpretation of copyright law is "inconsistent" with established legal doctrines. The lawyers warned that a ruling in favor of the media giant could have "disastrous consequences" for the American economy. "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 letter stated.
Chronology of the Legal Conflict
The legal battle began in late 2023 when the New York Times filed a comprehensive lawsuit against OpenAI and its primary financial backer, Microsoft. The lawsuit alleged that the companies had utilized millions of the paper’s articles to train their AI models, including ChatGPT, without obtaining permission or providing compensation. The Times argued that this practice constituted a massive "free ride" on its multi-billion-dollar investment in journalism.
Following the initial filing, OpenAI responded by claiming that the Times had "hacked" its system by using highly specific prompts to force the AI to regurgitate snippets of copyrighted text—a phenomenon known as "memorization." OpenAI argued that such instances are rare and do not represent the typical functioning of the model.
Throughout 2024, the case became a focal point for the broader debate over "AI ethics" and "creator rights." Dozens of other lawsuits followed, involving authors like George R.R. Martin, visual artists, and music publishers. The intervention by the Trump administration this week represents the most significant political involvement in the case to date, effectively elevating the dispute from a private commercial quarrel to a matter of national economic policy.
Supporting Data: The Economic and Technological Stakes
The government’s intervention is backed by the immense economic projections surrounding generative AI. Industry analysts at Goldman Sachs have estimated that generative AI could drive a 7% (or nearly $7 trillion) increase in global GDP over a ten-year period. For the United States, maintaining a competitive edge against rivals like China is a matter of both economic and national security.
The dataset in question is equally massive. OpenAI’s GPT-4 is believed to have been trained on trillions of "tokens," or pieces of words. The New York Times’ archive, spanning over 170 years, represents one of the highest-quality sources of English-language text available for such training. High-quality data is essential for reducing "hallucinations" (instances where AI generates false information) and improving the reasoning capabilities of LLMs.
The government’s letter also addressed the competitive impact of AI on traditional media. One of the four factors of the "fair use" test is the effect of the use upon the potential market for or value of the copyrighted work. The administration argued that OpenAI’s models do not "significantly compete" with the New York Times’ articles. While an AI might summarize a news event, the government contends that this is not a market substitute for the experience of reading original journalism or maintaining a subscription to a news outlet.
The Philosophy of Learning: Humans vs. Machines
One of the more striking arguments in the government’s brief was a comparison between AI training and human learning. The lawyers invoked a famous anecdote involving the late author Joan Didion, who reportedly typed out Ernest Hemingway’s stories as a teenager to learn the rhythm and structure of his sentences. The government argued that if it is legal for a human writer to "train" on the works of others to master their craft, it should be similarly legal for a machine to do so.
This analogy seeks to debunk the idea that AI training is a form of mechanical copying. Instead, the administration frames it as an intellectual process of pattern recognition. If the court accepts this logic, it would create a formidable barrier for any plaintiff seeking to claim copyright infringement based solely on the training phase of AI development.
Precedents and Parallel Litigations
The New York Times case does not exist in a vacuum. Several other rulings have recently shaped the legal landscape for AI:
- Meta (Kadrey v. Meta): In 2023, Meta secured a partial victory in a copyright case brought by authors. The judge noted that the plaintiffs failed to prove that the outputs of Meta’s LLaMA model were "substantially similar" to their copyrighted books. However, the court left the door open for future claims if training could be proven to cause specific harm.
- Anthropic ($1.5 Billion Settlement): In a contrasting outcome, the AI startup Anthropic recently lost a major case and was ordered to pay $1.5 billion in damages to a group of authors—the largest copyright settlement in U.S. history. Notably, the judge in that case ruled that while the training itself might be fair use, the company had essentially "pirated" the books by obtaining them from illegal shadow libraries, thus violating the law before the training even began.
- The Music Industry: Sony Music and Warner Music recently sued Anthropic, following a similar suit from Universal Music Group. These cases allege that AI models are being trained to replicate the distinct styles of artists, potentially infringing on both copyright and "right of publicity" laws.
Official Responses and Stakeholder Reactions
The reaction to the government’s letter was swift and divided. The New York Times expressed sharp disapproval, framing the administration’s stance as an abandonment of individual creators in favor of corporate giants.
"The Administration is siding with a handful of trillion-dollar AI companies at the expense of the countless American creators whose work they stole," said Graham James, a spokesperson for the New York Times. "Both AI and creators can thrive—AI companies simply need to pay fairly for the content that makes their products possible, as copyright law requires."
Mary Rasenberger, CEO of the Authors Guild—which is currently engaged in its own litigation against OpenAI—echoed these sentiments. She described the government’s letter as "replete with faulty arguments and a gross misunderstanding of the fair use doctrine."
On the other hand, legal experts noted the strategic weight of the Department of Justice’s involvement. Evan Brown, an intellectual property lawyer at Neal & McDevitt, told reporters that while District Judge Sidney H. Stein is not legally bound by the DOJ’s letter, such submissions "inherently carry a lot of weight" and are likely to be treated with great seriousness by the judiciary.
Pamela Samuelson, a copyright expert at the Berkeley Center For Law & Technology, characterized the government’s brief as a "significant development" that aligns with the majority of AI-related copyright rulings seen thus far, which have generally favored the "transformative" nature of the technology.
Broader Impact and Implications for the Future
The intervention by the Trump administration sets the stage for a potentially definitive ruling on the future of intellectual property in the digital age. If the court adopts the government’s view, it will establish a "safe harbor" for AI companies to use publicly available data for training purposes without the need for individual licenses. This would likely accelerate the development of AI tools but could lead to a financial crisis for content-heavy industries like journalism, publishing, and the arts.
Conversely, if the New York Times prevails, it could mandate a licensing regime that would cost the AI industry billions of dollars. While this would provide a new revenue stream for creators, some tech advocates argue it would create a "pay-to-play" environment where only the wealthiest companies—such as Microsoft, Google, and Meta—could afford to build advanced AI, effectively stifling competition from smaller startups.
As the case moves toward a potential trial or summary judgment, the legal community is watching closely for how Judge Stein will weigh the government’s national interest arguments against the foundational protections of the Copyright Act. The outcome will likely determine whether the United States maintains a "permission-less" innovation model for AI or shifts toward a more regulated framework that prioritizes the rights of original content producers. For now, the federal government has made its choice clear: in the race for AI supremacy, the speed of innovation takes precedence over the traditional boundaries of copyright.
About the Author
