Trump Administration Backs OpenAI Against New York Times in Landmark AI Copyright Fight

AI, Copyright and the Fight for America’s Technological Future

The Court Case That Could Reshape Artificial Intelligence

The Trump administration has thrown the weight of the United States government behind OpenAI in one of the most important copyright battles of the artificial intelligence era, arguing that training large language models on copyrighted text can qualify as fair use. The intervention pushes a dispute that began as a fight between technology companies and publishers into something far bigger: a battle over America's technological power, the future economics of human-created content and the legal foundations on which generative AI has been built.

The Justice Department filed a statement of interest on September 1 in the consolidated OpenAI copyright litigation before the US District Court for the Southern District of New York. Its message was unusually clear. Washington believes the court should reject the proposition that training large language models on copyrighted text is, by itself, a violation of copyright law.

Washington Has Chosen a Side on the Central AI Question

The case grew out of litigation launched by the New York Times against OpenAI and Microsoft in December 2023. The publisher alleges that enormous quantities of its copyrighted journalism were used without permission to help develop generative AI systems and argues that those systems can compete with the very journalism from which they learned.

OpenAI has pushed a fundamentally different interpretation. Its position rests heavily on fair use, the doctrine that permits certain uses of copyrighted works without the copyright holder's permission depending on factors including purpose, transformation, the amount used and the effect on the market for the original work.

The federal government has now moved firmly towards that side of the argument on AI training. Its intervention does not automatically give OpenAI victory, and the government is not replacing the judge or deciding the underlying factual disputes. It does, however, mean that one of the most consequential arguments advanced by AI developers now carries the explicit support of the US government.

That matters because Washington is not describing the dispute merely as a private commercial argument over licensing fees. The administration is connecting access to training material directly to America's economic competitiveness, scientific development and national security.

The Government Says AI Training Is Transformative

At the centre of the Justice Department's position is the claim that large-scale AI training is highly transformative. An AI model does not simply operate as a conventional archive in which a reader retrieves an unchanged article from storage. During training, huge quantities of material are analysed statistically so that a model can learn patterns and relationships and generate new responses.

That distinction could become critical. Copyright law has long wrestled with situations in which copyrighted material is copied as an intermediate step towards creating a substantially different function. The question now facing the courts is whether training modern generative AI systems fits comfortably within those principles or whether the scale, commercial value and output capabilities of the technology fundamentally change the calculation.

The administration argues that restricting this process could impose consequences far beyond individual technology companies. Large language models are increasingly being used for scientific research, software development, education, medicine, business analysis and other fields. Washington's position is that interpreting copyright law in a way that obstructs the development of those systems could ultimately weaken American innovation.

That argument fits directly into Donald Trump's wider AI policy. His administration has repeatedly framed leadership in artificial intelligence as a strategic contest in which the United States cannot afford to surrender technological ground to foreign competitors.

The New York Times Says Creators Would Pay the Price

The counterargument is powerful because the technology does not exist in an economic vacuum. Books, photographs, reporting, music, software and other creative works can cost enormous amounts of money to produce, while training an AI system on those works can potentially occur without a direct payment to the people who created them.

The New York Times argues that AI companies should pay fairly for content used to build commercially valuable systems. Its broader concern is that allowing technology companies to consume copyrighted material without permission or compensation could weaken the financial model supporting the production of original human-created material.

That exposes the central contradiction at the heart of the entire AI copyright war. Artificial intelligence becomes more capable when it can learn from enormous amounts of high-quality human work. Yet if those systems eventually reduce demand, traffic or revenue for the organisations and individuals producing that work, the technology could damage part of the ecosystem on which it depends.

The argument therefore extends well beyond whether a particular computer copied a particular article at a particular moment. Courts are effectively being asked to determine how copyright law designed long before generative AI should divide economic power between those who create information and those who build machines capable of learning from it.

A Mandatory Licensing World Could Have Its Own Winners

There is another side to the economics. If every AI developer needed to negotiate licences covering enormous quantities of training material before building a competitive model, the costs could be immense.

That could strengthen copyright holders, but it could also reinforce the dominance of the richest technology companies. A corporation capable of spending billions of dollars acquiring data and negotiating licensing agreements may survive a compulsory licensing environment relatively comfortably. A university laboratory, startup or independent AI developer may not.

This is one reason the government's argument reaches beyond OpenAI itself. A legal regime intended to restrain the largest AI corporations could inadvertently construct an enormous financial barrier protecting those same corporations from smaller competitors.

The result would be deeply ironic. Copyright restrictions designed partly to stop technology giants accumulating too much power could make frontier AI development so expensive that only technology giants could afford to participate.

The Fight Is Not Simply Copyright Holders Against Technology

The dividing line is already becoming more complicated than a straightforward battle between publishers and AI companies. Some media and technology businesses have negotiated licensing arrangements rather than relying entirely on litigation. Others have chosen to fight.

That produces a second major question for courts and lawmakers. If AI training ultimately qualifies broadly as fair use, licensing copyrighted material may become a commercial choice rather than a legal necessity. If courts rule the opposite way, access to enormous libraries of copyrighted information could become one of the most expensive inputs in the AI economy.

Either outcome could redistribute billions of dollars.

A publisher-friendly interpretation could create a new global licensing market in which creators, archives, publishers and other rights holders sell access to their material for model development. A broad victory for AI developers could instead establish that much of the information publicly accessible to humans can also be computationally analysed by machines without a separate payment every time it contributes to training.

Those are radically different futures.

The Case Could Shape Far More Than Journalism

Although the New York Times litigation has become one of the highest-profile examples, the underlying legal argument reaches books, photographs, computer code, visual art, music and potentially almost every other category of copyrighted digital material.

AI developers need staggering quantities of information. The modern generative AI boom was made possible partly because decades of human knowledge, conversation and creative work had already migrated onto digital networks.

Copyright holders increasingly argue that accessibility is not the same thing as permission. Something being obtainable through the internet does not automatically eliminate the rights attached to it.

AI companies counter that learning from information has always involved exposure to existing work and that copyright does not ordinarily grant an author ownership over facts, ideas, styles, concepts or knowledge extracted from that work.

Generative AI forces those two principles into direct collision at unprecedented scale.

Trump Has Turned the Argument Into a Question of National Power

The administration's intervention also reveals how dramatically the political perception of artificial intelligence has shifted. AI is no longer treated purely as another technology industry requiring regulation.

Washington increasingly treats advanced AI infrastructure, models and computing capacity as strategic assets.

That means copyright policy is becoming entangled with geopolitical competition. If American developers face expensive or restrictive training rules while foreign competitors gain easier access to enormous datasets, the administration fears the United States could lose ground in a technology expected to influence military capability, scientific discovery, intelligence, manufacturing and economic productivity.

Associate Attorney General Stanley Woodward has framed AI dominance as important to national security, prosperity and economic mobility. Commerce Secretary Howard Lutnick has similarly encouraged an international approach that recognises fair-use principles while still seeking ways to protect creators.

The administration is therefore attempting to hold two positions simultaneously: copyrighted creative work deserves protection, but the law should not become an obstacle capable of choking American AI development.

Whether courts can produce a doctrine that genuinely achieves both objectives is another matter.

Fair Use Is Not a Blank Cheque

One important distinction risks being lost in the political rhetoric surrounding the case. A ruling that training can constitute fair use would not necessarily mean AI companies gain unlimited permission to do anything they want with copyrighted material.

Training and output are separate legal questions.

A system analysing copyrighted works during development may present a different copyright problem from a system subsequently producing substantial passages that closely replicate those works. How material was obtained could also matter, as could the nature of individual datasets and the effect a particular AI product has on the market for the underlying work.

That is why sweeping claims that the government has simply declared AI exempt from copyright would go too far. It has not.

Instead, Washington has intervened forcefully on the foundational training question and told the court that the transformative character and wider societal benefits of AI should carry enormous weight.

The Government Cannot Decide the Case

The Justice Department's filing is important, but it is not a judgment.

The federal government has submitted its interpretation of the law because it believes major national interests are involved. The court remains responsible for applying copyright doctrine to the evidence and arguments before it.

That distinction makes the next stage particularly important. A judicial ruling favouring OpenAI could become a powerful reference point for AI companies confronting similar claims across the United States. A ruling favouring copyright holders could accelerate demands for licensing agreements and force developers to reconsider how training datasets are assembled.

Higher courts may ultimately be required to settle some of these issues regardless of the initial result. Congress could also intervene if lawmakers conclude that copyright legislation written for an earlier technological era cannot adequately resolve the economics of generative AI.

The Real Question Is Who Pays for the AI Revolution

The easiest way to describe this lawsuit is as OpenAI against the New York Times. The deeper fight is over who captures the value created when machines learn from generations of human work.

Creators argue that powerful corporations should not be able to build enormously valuable products from copyrighted material while refusing to compensate the people who produced it. AI developers argue that forcing licences for the basic act of computational learning could cripple innovation, favour companies rich enough to buy access and allow copyright ownership to restrict the creation of genuinely new technology.

The Trump administration has now made clear where it stands on that central training question.

Its intervention significantly raises the stakes without settling the law. If the courts ultimately accept the government's interpretation, the decision could give American AI developers considerably greater freedom to train future models. If copyright holders prevail instead, one of the fundamental economic assumptions behind the generative AI boom may have to be rewritten.

The outcome will therefore matter far beyond one chatbot or one publisher. Courts are beginning to decide whether the vast digital record created by human beings is primarily protected inventory that machines must pay to learn from, or part of the informational environment from which a new generation of technology can lawfully learn. Whichever answer survives the coming appeals could shape artificial intelligence for decades.