{"slug": "trump-doj-backs-openai-in-new-york-times-copyright-case", "title": "Trump DOJ backs OpenAI in New York Times copyright case", "summary": "The Trump administration's Justice Department filed a 20-page statement of interest on September 1st in the Southern District of New York, backing OpenAI in its copyright dispute with The New York Times by arguing that training large language models on copyrighted text is transformative fair use and should be analyzed separately from potentially infringing outputs. The government also warned that forcing developers to license every training work would entrench large AI labs and publishers, weakening U.S. competition.", "body_md": "# Trump DOJ backs OpenAI in New York Times copyright case\n\n**The government says licensing requirements would protect large publishers, entrench rich AI labs and weaken U.S. competition.**\n\nBy [Ryan Merket](/author/ryan-merket)\n· Published\n\nPrimary source: [TechCrunch](https://techcrunch.com/2026/09/02/u-s-government-sides-with-openai-on-issue-of-training-llms-on-copyrighted-material/)\n\n## Why it matters\n\nThe filing gives OpenAI federal backing for separating model training from infringing outputs, a distinction that could determine who can afford to build U.S. foundation models.\n\nThe Trump administration has asked a federal judge to reject broad copyright liability for training large language models on protected text, putting the U.S. government behind [OpenAI (@OpenAI)](https://x.com/OpenAI) on the central issue in its legal fight with The New York Times.\n\nThe Justice Department filed its [20-page statement of interest](https://storage.courtlistener.com/recap/gov.uscourts.nysd.641355/gov.uscourts.nysd.641355.316.0.pdf) on September 1st in the consolidated OpenAI copyright litigation in the Southern District of New York. [TechCrunch reported](https://techcrunch.com/2026/09/02/u-s-government-sides-with-openai-on-issue-of-training-llms-on-copyrighted-material/) the filing on Wednesday.\n\nThe government argues that copying written works to train an LLM is highly transformative because the resulting model learns statistical relationships among words and can perform tasks different from those served by the original articles. It urged the court to analyze model training separately from any output that reproduces protected expression.\n\nThat distinction gives OpenAI the defense it has sought since the Times sued in December 2023. The Times alleges that OpenAI and Microsoft copied its journalism to build products that compete with the newspaper and can reproduce portions of its articles. OpenAI maintains that training is protected by fair use.\n\nThe Justice Department's filing does not decide the case, and the judge is not required to adopt its position. It gives OpenAI a powerful institutional ally, however, while recasting a commercial copyright dispute as a question of national security and industrial policy.\n\n### The government draws a line between training and outputs\n\nThe Justice Department says training an LLM requires copying works, converting the data into numerical representations and using those representations to learn linguistic patterns. In its view, that process serves a fundamentally different purpose from publishing an article for readers.\n\nThe filing concedes that an output may present a separate copyright problem when a model reconstructs and distributes protected material. The government nevertheless argues that potentially infringing outputs should not determine whether the earlier training process qualifies as fair use.\n\nThat separation matters because the Times has alleged that OpenAI models can generate portions of its journalism. The Justice Department's position would force publishers to prove infringement at the output level without treating the underlying act of training as categorically unlawful.\n\nA federal judge previously allowed key parts of the publishers' case to proceed. An [April 2025 opinion](https://www.nysd.uscourts.gov/sites/default/files/2025-04/yf%2023cv11195%20OpenAI%20MTD%20opinion%20april%204%202025.pdf) described separate acquisition, training and output stages, each of which can raise distinct copyright questions.\n\n### Licensing becomes a competition argument\n\nThe government's sharpest intervention concerns licensing. The Justice Department argues that forcing developers to pay for every work used in training could leave only the largest technology companies able to build competitive models. At the same time, publishers with the deepest archives would collect the largest payments.\n\nThe brief calls those licensing barriers a route to an oligopoly in model development and says they would function mainly as subsidies for established media companies. It also warns that U.S. developers would be placed at a disadvantage against foreign competitors operating under different intellectual-property rules.\n\nThat argument turns the publishers' demand for compensation into a market-structure question. A licensing requirement could create revenue for news organizations whose reporting supplies valuable training material. It could also raise the cost of building foundation models beyond the reach of smaller labs, further concentrating development among companies already able to afford massive computing and data bills.\n\nThe Justice Department did not ask the court to excuse every use of copyrighted material. Its filing says fair use remains fact-dependent and acknowledges that model outputs reproducing protected expression may require a different analysis. It argues that the court should reject a rule making unlicensed LLM training broadly impermissible.\n\n### The Copyright Office took a narrower view\n\nThe administration's position goes further than the U.S. Copyright Office's [May 2025 generative AI report](https://www.copyright.gov/ai/Copyright-and-Artificial-Intelligence-Part-3-Generative-AI-Training-Report-Pre-Publication-Version.pdf?stream=top). The Copyright Office concluded that some training uses would qualify as fair use and others would not, depending on the source material, purpose, output controls and market effects.\n\nThat report said noncommercial research with non-substitutive outputs was likely to receive fair-use protection. It found a weaker case where companies commercially exploit large collections of expressive works, especially material acquired illegally, to generate content competing in existing markets. The office also said developing licensing markets could count against unlicensed use under the market-harm factor.\n\nThe Justice Department directly attacks that market-harm theory when the alleged injury comes from generalized competition rather than outputs substantially similar to protected works. It argues that copyright does not grant writers ownership over a genre, facts or the ability to compete against new works created with a model.\n\nThe filing now gives the court two competing federal views. The Copyright Office offered a case-by-case framework that leaves room for licensing and liability. The Justice Department wants the court to establish a much stronger presumption that LLM training advances the constitutional purpose of copyright.\n\nFor OpenAI, that difference reaches beyond damages in one lawsuit. A ruling that treats training itself as infringement could make access to licensed text a core cost of model development. A fair-use ruling would preserve the industry's ability to ingest protected work first and litigate particular outputs later.", "url": "https://wpnews.pro/news/trump-doj-backs-openai-in-new-york-times-copyright-case", "canonical_source": "https://runtimewire.com/article/trump-doj-backs-openai-in-new-york-times-copyright-case", "published_at": "2026-09-02 17:34:52+00:00", "updated_at": "2026-09-02 17:54:56.012572+00:00", "lang": "en", "topics": ["ai-policy", "large-language-models", "ai-ethics"], "entities": ["OpenAI", "The New York Times", "Justice Department", "Trump administration", "Microsoft", "TechCrunch"], "alternates": {"html": "https://wpnews.pro/news/trump-doj-backs-openai-in-new-york-times-copyright-case", "markdown": "https://wpnews.pro/news/trump-doj-backs-openai-in-new-york-times-copyright-case.md", "text": "https://wpnews.pro/news/trump-doj-backs-openai-in-new-york-times-copyright-case.txt", "jsonld": "https://wpnews.pro/news/trump-doj-backs-openai-in-new-york-times-copyright-case.jsonld"}}