--- title: DOJ backs OpenAI's fair use defense in publisher... description: The US Department of Justice filed a statement of interest backing OpenAI's fair use defense in the consolidated publisher copyright lawsuits in New York,... canonical: https://daily.dev/posts/doj-backs-openai-s-fair-use-defense-in-publisher-copyright-lawsuits-bg2fpxouo twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: DOJ backs OpenAI's fair use defense in publisher copyright lawsuits | daily.dev og:description: The US Department of Justice filed a statement of interest backing OpenAI's fair use defense in the consolidated publisher copyright lawsuits in New York,... og:url: https://daily.dev/posts/doj-backs-openai-s-fair-use-defense-in-publisher-copyright-lawsuits-bg2fpxouo og:image: https://api.daily.dev/og/posts/Bg2fpXoUo.png og:image:alt: DOJ backs OpenAI's fair use defense in publisher copyright lawsuits og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # DOJ backs OpenAI's fair use defense in publisher copyright lawsuits **[Collections](https://daily.dev/sources/collections)** · 3 min read · 2 upvotes · 0 comments ## Summary The US Department of Justice filed a statement of interest backing OpenAI's fair use defense in the consolidated publisher copyright lawsuits in New York, arguing that training LLMs on copyrighted text is transformative and doesn't cause direct market harm. The filing warns that mandatory licensing would hurt smaller US AI companies against foreign rivals, framing it as a national security issue, while leaving questions about data acquisition and output reproduction unresolved. Meanwhile the EU takes an opposite approach: it lacks a fair use doctrine, relies on an opt-out based text and data mining exception, and the EU AI Act requires respecting rightsholder opt-outs regardless of where training happened, creating a widening transatlantic legal divide for AI companies. ## Content The US Justice Department has filed a statement of interest in the consolidated publisher copyright cases against OpenAI in New York, formally backing the argument that training large language models on copyrighted text qualifies as fair use. The New York Times, Ziff Davis, and hundreds of local newspapers are suing OpenAI over its use of their content to train AI models. This is the first time the federal government has formally intervened in the broader wave of AI copyright litigation, though the filing is advisory rather than binding on the court. ## What the DOJ actually argued The filing draws a clear line between three separate acts: acquiring training data, training on it, and generating outputs. The DOJ's argument focuses specifically on the training stage. On transformativeness, the filing is blunt: "The copying of protected text articles as part of training an LLM is a use of a different kind or character that is 'transformative—spectacularly so.'" The reasoning is that an LLM uses text to learn statistical relationships, not to reproduce or substitute for the original work. On market harm, the DOJ argues that training itself doesn't replace the original articles, so AI-generated competition downstream shouldn't automatically make the earlier training unlawful. The filing also warns that blanket licensing requirements would raise barriers for smaller AI companies and disadvantage US developers against foreign competitors, framing the issue explicitly as a national security concern: "Rules of law that make it significantly more difficult to develop a robust AI industry in the United States therefore threaten national security and give a competitive advantage to foreign adversaries." Importantly, the DOJ didn't give AI companies a blanket pass. Questions about how training data was acquired and whether specific outputs reproduce protected passages remain open. ## What this does and doesn't settle If the court adopts the DOJ's framework, it would shift legal pressure away from model training itself and toward data acquisition practices and specific outputs that closely reproduce source material. The court still has to evaluate fair use case by case. ## The EU sees this very differently The DOJ's national security argument has no equivalent in European copyright law. The EU has no fair use doctrine. Instead, it relies on a text and data mining exception under Article 4(3) of the 2019 copyright directive, which includes an explicit opt-out mechanism for rightsholders. The EU AI Act goes further: Article 53 and Recital 106 require any provider placing a model on the EU market to respect rightsholder opt-outs regardless of where training occurred. That directly contradicts the DOJ's position. A Munich court already ruled against OpenAI on similar grounds involving GPT-4 and song lyrics. That decision is under appeal and could eventually reach the Court of Justice of the EU. So while Washington is moving to protect AI training under fair use, Brussels is moving in the opposite direction. For companies operating in both markets, that tension isn't going away. ## Questions this post answers ### Did the US Justice Department say AI training on copyrighted articles is fair use? Yes, the DOJ filed a statement of interest in the consolidated New York publisher copyright cases against OpenAI, arguing that training an LLM on copyrighted text is transformative use because the model learns statistical relationships rather than reproducing the work. The filing is advisory, not binding, and doesn't address data acquisition or output reproduction, which remain open legal questions. _developers weighing legal exposure from AI training data can follow how this fair use debate unfolds on daily.dev._ ### How does the EU's approach to AI training data copyright differ from the US fair use argument? The EU has no fair use doctrine and instead relies on a text and data mining exception under Article 4(3) of its 2019 copyright directive, which lets rightsholders opt out of having their content used for training. The EU AI Act's Article 53 and Recital 106 require any provider placing a model on the EU market to honor those opt-outs regardless of where training took place, contradicting the DOJ's fair use stance. _teams shipping AI models across US and EU markets can track this regulatory divide on daily.dev._ ## Community take How the wider developer community reacted, aggregated from 4 discussions and 34 comments across x (as of 2026-09-04). **TL;DR:** Reactions are split between people who see the DOJ filing as a sensible, expected legal stance on AI training, and others who view it as government favoritism toward AI companies at the expense of creators and copyright law. **Sentiment:** 30% positive · 25% mixed · 45% skeptical **The case for** - Some argue training on text is analogous to a person reading and drawing inspiration from books, so it shouldn't be treated as theft. - A few see the filing as a rare instance of government taking a reasonable, pragmatic legal position. - Some frame licensed data provenance as a coming competitive/business feature rather than a legal blocker. **The pushback** - Many argue the fair-use question on training is separate from the regurgitation/output issue, so a win on one doesn't resolve the other. - Some see the DOJ's involvement as evidence the government prioritizes AI industry interests over journalists and copyright holders. - Others call the national-security framing narrow and premature to treat as a general precedent for all AI training use cases. - A vocal minority frames this as legalized theft of creative and journalistic work benefiting large, well-connected companies. **By community** - x (heated): Replies swing sharply between framing the DOJ stance as overdue common sense and as government-sanctioned theft of copyrighted work, with a long combative sub-thread arguing over whether training and verbatim output are the same issue. **Hottest debate:** Whether training an LLM on copyrighted text is legally and morally equivalent to human learning/inspiration, or whether it constitutes unauthorized copying and derivative-work infringement. **Open questions** - Does a fair-use win on training data reach or resolve the separate question of verbatim output reproduction? - Will the broad DOJ position translate into consistent rulings across the many other pending AI copyright lawsuits (Anthropic, Meta, etc.)? - How will provenance of training data (licensed vs. paywalled/scraped) be verified or enforced going forward? **Highlights** > @rohanpaul_ai the two claims are separate though. whether training on the text is fair use is a different question from whether the model spits chunks of it back out. a win on the first doesnt reach the second, and the times complaint leaned hard on the regurgitation examples > — [AIQuanting on x · 7 points](https://x.com/AIQuanting/status/2095198222436479267) > @rohanpaul_ai The US government is basically stating it does not care about copyright laws because it wants to be able to use the stolen data via LLMs. The New York Times is correct in claiming the model uses the stolen data to create derivative works, and/or competing products. > — [HowardAulsbrook on x · 1 comments](https://x.com/HowardAulsbrook/status/2095220969061138663) > @HowardAulsbrook @rohanpaul_ai So, if a human gains inspiration from a book or some other form of art, it should be against copyright law for that human to use that inspiration to create their own book or art? > — [CIA\_DD on x · 3 points, 1 comments](https://x.com/CIA_DD/status/2095226375007309869) > @rohanpaul_ai calling it a win for AI training in general is premature. the government backed it on national security grounds, that's an incredibly narrow argument. that precedent doesn't help a startup training on newspaper archives for a recipe app > — [Chahatusharma on x · 1 points](https://x.com/Chahatusharma/status/2095197239459061820) > @rohanpaul_ai massive win for looters of all kinds! first steps in a worldwide OFFICIAL GOVERNMENTAL SANCTIONING, nay, ENCOURAGEMENT of all thugs & gangs & warlords! nobody, no group, no industry, need ever pay for raw materials if they can get them through chicanery, stealth, bribery & force. > — [artCarnivore on x](https://x.com/artCarnivore/status/2095715227199242299) **Source threads** - [x](https://x.com/tobi/status/2095301343720222956) · 0 points · 0 comments - [x](https://x.com/rohanpaul_ai/status/2095196217634640144) · 0 points · 13 comments - [x](https://x.com/rohanpaul_ai/status/2095197606884573208) · 0 points · 21 comments - [x](https://x.com/rohanpaul_ai/status/2095490988097900874) · 0 points · 0 comments ## Similar posts on daily.dev - [News outlets ask judge to sanction OpenAI in copyright case](https://daily.dev/posts/news-outlets-ask-judge-to-sanction-openai-in-copyright-case-gddcthoev) · The Next Web · 0 upvotes · 0 comments --- Tags: [#openai](https://daily.dev/tags/openai), [#ai-governance](https://daily.dev/tags/ai-governance), [#ai-regulation](https://daily.dev/tags/ai-regulation) [View this post on daily.dev](https://daily.dev/posts/doj-backs-openai-s-fair-use-defense-in-publisher-copyright-lawsuits-bg2fpxouo) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"DOJ backs OpenAI's fair use defense in publisher copyright lawsuits","url":"https://daily.dev/posts/doj-backs-openai-s-fair-use-defense-in-publisher-copyright-lawsuits-bg2fpxouo","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/doj-backs-openai-s-fair-use-defense-in-publisher-copyright-lawsuits-bg2fpxouo"},"datePublished":"2026-09-03T12:36:21.723Z","dateModified":"2026-09-04T05:54:00.106Z","description":"The US Department of Justice filed a statement of interest backing OpenAI's fair use defense in the consolidated publisher copyright lawsuits in New York,...","isAccessibleForFree":true,"articleSection":"Collections","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Collections","logo":"https://media.daily.dev/image/upload/s--fk_6ycEi--/f_auto,q_auto/v1780996001/logos/collections?_a=BAMAMiWQ0","url":"https://daily.dev/sources/collections"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/doj-backs-openai-s-fair-use-defense-in-publisher-copyright-lawsuits-bg2fpxouo","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":2},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"openai,ai-governance,ai-regulation","timeRequired":"PT3M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Collections","item":"https://daily.dev/sources/collections"},{"@type":"ListItem","position":3,"name":"DOJ backs OpenAI's fair use defense in publisher copyright lawsuits"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/doj-backs-openai-s-fair-use-defense-in-publisher-copyright-lawsuits-bg2fpxouo#faq","mainEntity":[{"@type":"Question","name":"Did the US Justice Department say AI training on copyrighted articles is fair use?","acceptedAnswer":{"@type":"Answer","text":"Yes, the DOJ filed a statement of interest in the consolidated New York publisher copyright cases against OpenAI, arguing that training an LLM on copyrighted text is transformative use because the model learns statistical relationships rather than reproducing the work. The filing is advisory, not binding, and doesn't address data acquisition or output reproduction, which remain open legal questions. developers weighing legal exposure from AI training data can follow how this fair use debate unfolds on daily.dev."}},{"@type":"Question","name":"How does the EU's approach to AI training data copyright differ from the US fair use argument?","acceptedAnswer":{"@type":"Answer","text":"The EU has no fair use doctrine and instead relies on a text and data mining exception under Article 4(3) of its 2019 copyright directive, which lets rightsholders opt out of having their content used for training. The EU AI Act's Article 53 and Recital 106 require any provider placing a model on the EU market to honor those opt-outs regardless of where training took place, contradicting the DOJ's fair use stance. teams shipping AI models across US and EU markets can track this regulatory divide on daily.dev."}}]} ```