The Trump administration is supporting OpenAI in the NYT copyright lawsuit
The Trump administration has intervened in The New York Times’ copyright lawsuit against OpenAI, filing a statement of interest that backs the AI lab’s core defense: that training large language models on copyrighted text constitutes fair use. The move injects the weight of the federal government into a case that is widely viewed as the most significant legal test yet for the generative AI industry, signaling that the outcome could reshape the legal boundaries of how AI systems are built.
Anthropic launches Cowork, a Claude Desktop agent that works… OpenAI ends Microsoft legal peril over its $50B Amazon… Snap spins off AI video team into new company,… An Anthropic researcher just gave us a peek at…
Background
The lawsuit, filed in federal court in December 2023, alleges that OpenAI and its primary backer, Microsoft, unlawfully copied millions of articles published by The New York Times to train their AI systems. The Times argues that this ingestion of copyrighted material—performed without permission or payment—violates intellectual property law. The publisher is seeking "billions of dollars" in statutory and actual damages and has asked the court to order the destruction of any models or training sets that incorporate its work.
OpenAI has countered that its training process is protected by the fair use doctrine. The company argues that its models do not merely copy or regurgitate Times content but instead analyze patterns across a vast corpus of text to learn language, reasoning, and world knowledge. In its view, this transformative use—turning expressive works into a functional, non-expressive tool—falls squarely within established fair use precedent.
This week, the Department of Justice filed a statement of interest supporting that argument. While such filings do not decide the case, they signal the executive branch’s interpretation of the law to the court. The administration’s brief contends that The Times is attempting to narrow the fair use doctrine in a way that would categorically exclude the training of large language models, a move the government argues would be inconsistent with the statute’s flexible, case-by-case framework.
Why It Matters
The stakes extend far beyond the two parties in the courtroom. This case has become the de facto bellwether for a wave of similar litigation brought by authors, visual artists, music publishers, and other rights holders against AI developers. A ruling in favor of The Times could establish a precedent requiring AI companies to license all copyrighted training data—a requirement that would fundamentally alter the economics and feasibility of building frontier models. Conversely, a ruling affirming fair use for training would provide legal certainty for current development practices, though it would leave open questions about output similarity and market harm.
For businesses adopting or building AI tools, the intervention adds a new variable to risk assessments. Companies fine-tuning models on proprietary or third-party data, or deploying retrieval-augmented generation (RAG) systems that surface copyrighted text, are watching closely. The administration’s stance suggests a federal policy preference for innovation access over restrictive licensing regimes, but a court ruling could still impose compliance costs, data auditing requirements, or restrictions on model outputs.
The filing also highlights a tension in current copyright law: the fair use statute was written long before machine learning at scale existed. Courts must now decide whether "transformative use" covers the ingestion of entire works for statistical modeling, and whether the commercial nature of AI products weighs against that defense. The administration’s argument essentially asks the court to apply existing precedent flexibly rather than create a new categorical exception—or prohibition—for AI.
What’s Next
The case is currently in the discovery and motion phase in the Southern District of New York. Both sides have filed motions for summary judgment on various claims, and the court has not yet set a trial date. The statement of interest will be considered alongside amicus briefs from other stakeholders, including technology companies, library associations, and creator groups.
Meanwhile, the legislative branch is also circling the issue. Lawmakers have held hearings on AI and copyright, and several bills have been proposed to clarify training data transparency, opt-out mechanisms, or compulsory licensing frameworks. The judicial outcome in *New York Times v. OpenAI* will likely influence the urgency and shape of any federal legislation.
For OpenAI and Microsoft, the administration’s support is a meaningful endorsement, but it does not guarantee victory. The court will conduct its own four-factor fair use analysis, weighing the purpose and character of the use, the nature of the copyrighted work, the amount used, and the effect on the market. The Times has presented evidence of near-verbatim outputs from its articles, which could weigh heavily on the fourth factor.
Takeaway for Business Owners
The federal government has signaled it views AI training as fair use, but the courts have not decided. Until there is a final ruling—or legislation—businesses should treat training data provenance as a live compliance risk. Document your data sources, evaluate licensing where feasible, and monitor this case closely. The legal framework for generative AI is being written in real time, and the *Times v. OpenAI* decision will be a foundational precedent for years to come.
Source: Original Article