DOJ backs OpenAI fair use defense in NYT copyright suit

The intersection of artificial intelligence and copyright law has become a primary venue for defining the boundaries of digital reproduction. Historically, fair use jurisprudence has evolved through cases involving search engine indexing, text mining, and digital archiving, where courts often distinguished between copying for expression versus copying for functional analysis. The current controversy follows precedents set in Authors Guild v. Google (2015), where digitizing books for search functionality was deemed transformative. However, generative AI introduces novel variables: models produce competitive synthetic content rather than mere indices, complicating the fourth fair use factor regarding market harm. Previous disputes over web scraping and dataset creation established that non-expressive intermediate copying can be lawful, yet those cases rarely involved direct commercial competition between the copier and the copied. The pattern suggests a judicial tendency to favor technological utility when output does not substitute for input, though generative capabilities challenge this distinction. Regulatory bodies globally are currently attempting to codify exceptions for text and data mining, creating a fragmented landscape where U.S. fair use flexibility contrasts with stricter opt-in regimes elsewhere. This case tests whether existing flexible standards can accommodate probabilistic model training without legislative amendment.

On September 2, 2026, the U.S. Department of Justice (DOJ) formally intervened in the copyright infringement lawsuit filed by The New York Times against OpenAI, submitting a statement of interest that explicitly supports OpenAI’s fair use defense. According to reporting by The New York Times, the Justice Department sided with the AI developer in the ongoing litigation, marking a significant federal endorsement of the legal theory that training large language models on copyrighted material constitutes permissible use under U.S. law [3]. This filing represents the culmination of a legal dispute that began on December 27, 2023, when The New York Times filed a federal lawsuit alleging that OpenAI had unauthorizedly appropriated millions of copyrighted articles to train its generative AI products without licensing or compensation.

The DOJ’s intervention aligns with broader executive branch efforts to secure American technological competitiveness. According to a post by Wall Street Engine, Commerce Secretary Howard Lutnick urged G20 nations to embrace "fair use" standards that allow AI companies to train on creators' work while maintaining protections for artists and inventors [1]. The same source notes that the DOJ separately backed OpenAI in its specific copyright fight with the NYT, arguing that AI training generally qualifies as fair use [1]. This diplomatic and legal coordination suggests the administration views the outcome of this specific case as having implications beyond domestic intellectual property law, extending into international trade and technology policy.

Legal scholars have noted the specific doctrinal arguments employed by the government. Anupam Chander stated that the DOJ’s filing strongly argues that the reading of copyrighted works by Large Language Models (LLMs) for training purposes is fair use [2]. Furthermore, according to Chander, the DOJ explicitly rejected the Copyright Office's pre-publication theory of "market dilution" regarding AI outputs, citing legal scholarship in support of this position [2]. This rejection of market dilution theories is a critical component of the defense, as it addresses concerns that AI models might supplant the original market for news content even if they do not reproduce it verbatim.

OpenAI has consistently maintained that its training practices are protected. On June 1, 2024, the company filed a fair use defense motion arguing that training constitutes transformative use protected under copyright doctrine. The company’s position rests on the premise that LLMs learn patterns and linguistic relationships rather than copying expressive content, a distinction central to the transformative use analysis. The DOJ’s backing validates this technical-legal interpretation at the federal level, potentially insulating AI developers from mass litigation if courts adopt similar reasoning.

Conversely, The New York Times maintains that the unauthorized ingestion of its proprietary journalism represents an existential threat to the news industry. The publisher’s original complaint alleged that OpenAI built commercial products directly competing with the sources that trained them, bypassing established licensing markets. While the provided sources focus heavily on the government’s defense, the underlying controversy hinges on whether AI training is truly transformative or merely a high-tech form of free-riding. Jensen Huang, referenced in the context of the broader policy debate, urged policymakers to regulate "real-world problems" rather than "theoretical harms," signaling industry frustration with precautionary approaches to IP enforcement [1].

The sequence of events demonstrates a tightening alignment between federal policy and AI industry interests. From the initial lawsuit in late 2023 to OpenAI’s affirmative defense in mid-2024, and finally to the DOJ’s statement of interest in September 2026, the trajectory shows increasing institutional support for the fair use argument. The involvement of the Commerce Department in parallel G20 discussions indicates that the administration treats this legal question as a matter of national economic strategy. However, despite this executive backing, the ultimate determination remains with the judiciary, which must balance statutory copyright protections against the government’s articulated innovation priorities.

The New York Times faces potential permanent devaluation of its archival content if AI training is deemed fair use, threatening subscription and licensing revenue streams derived from millions of articles. OpenAI and the broader generative AI sector stand to gain a durable liability shield that could reduce compliance costs and accelerate model development. The U.S. government risks diplomatic friction with allies enforcing stricter copyright regimes while betting that domestic AI leadership outweighs creator economy harms. Magnitude encompasses the entire news publishing industry's digital transition and the foundational business model of Western AI companies.