{"id":6986,"date":"2026-10-02T12:47:34","date_gmt":"2026-10-02T12:47:34","guid":{"rendered":"https:\/\/blogs.luc.edu\/compliance\/?p=6986"},"modified":"2026-10-02T12:47:34","modified_gmt":"2026-10-02T12:47:34","slug":"ai-training-and-copyright-navigating-compliance-while-the-rules-remain-unclear","status":"publish","type":"post","link":"https:\/\/blogs.luc.edu\/compliance\/?p=6986","title":{"rendered":"AI Training and Copyright: Navigating Compliance While the Rules Remain Unclear"},"content":{"rendered":"

Esther Egunyomi<\/em><\/p>\n

Associate Editor<\/em><\/p>\n

Loyola University Chicago School of Law, JD 2028<\/em><\/p>\n

Generative artificial intelligence (AI) has quickly<\/a> become part of everyday business practices, but the technology presents an unresolved question for copyright law: can companies use copyrighted material to train artificial intelligence models without acquiring permission from copyright owners? That question is now central to major litigation<\/a> involving OpenAI, Microsoft, The<\/em> New York Times, <\/em>authors, and other new organizations. As courts determine how traditional copyright principles apply to AI training, companies face a compliance challenge: create innovative AI systems while mitigating the risk of copyright infringement.<\/p>\n

Copyright law meets AI training<\/strong><\/p>\n

Copyright law<\/a> provides creators with a limited set of exclusive rights in their original works, which are further constrained by certain legislative enactments, such as the fair use doctrine. Section 107 of the Copyright Act defines<\/a> fair use as the \u201cuse of a copyrighted work for purposes such as criticism, comment, news reporting, teaching (including multiple copies for classroom use), scholarship, or research,\u201d regardless of whether such use is for commercial or nonprofit purposes. When<\/a> deciding whether a use qualifies as fair use, courts consider several factors, including the nature of the copyrighted work, the amount and substantially of the portion used in relation to the copyrighted work, and the effect of the use on the potential market for or value of the copyrighted work. The Fair Use Index of the U.S. Copyright Office<\/a> emphasizes that fair use is determined on a case-by-case basis and requires consideration of the particular circumstances rather than application of a single formula.<\/p>\n

AI complicates the fair use analysis<\/a> because developers train models using enormous quantities of text, images, and other materials, some of which is protected by copyright. In its report, Copyright and Artificial Intelligence, Part 3: Generative AI Trainin<\/em>g<\/a>, the Copyright Office concluded that the fair use analysis for AI training, like other forms of copyright usage, cannot be reduced to a simple rule. The Copyright Office<\/a> explained that whether AI training qualifies as fair use depends on a list of factors, including the works used, their source, the purpose of the training, and the effect of the resulting model on markets for copyrighted works.<\/p>\n

The New York Times litigation raises the stakes<\/strong><\/p>\n

The right to use copyright protected materials to train AI models under fair use doctrine is currently being tested in federal court. The New York Times<\/em><\/a> and other copyright owners have accused OpenAI and Microsoft of using copyrighted material, without authorization, while developing AI systems. This litigation<\/a> is now at a particularly important juncture: in September 2026, the parties filed arguments on summary judgment as to whether the challenged uses fall under the protection of the fair use doctrine. Some of the dispute hinges on whether the training of AI sufficiently transforms copyrighted materials and whether the products of AI can replace original works, harming their value.<\/p>\n

The lawsuit underscores the unsettled legal landscape. Some earlier federal district court rulings relating to Anthropic and Meta found certain<\/a> uses of copyrighted books in AI training to be fair use. However, those rulings did not establish a universal precedent protecting AI training. Questions involving pirated copies<\/a>, market harm, substitution, and reproduction of copyrighted materials remain unresolved. Although, the Copyright Office has been examining these questions through its broader Copyright and Artificial Intelligence Initiative<\/a>. \u00a0In May 2025, the Copyright Office released a pre-publication version of its report addressing copyright issues raised by AI training after receiving extensive public<\/a>input, and it identified AI training as an area involving unresolved copyright and policy questions.<\/p>\n

Compliance cannot wait for the courts<\/strong><\/p>\n

The courts may eventually provide further clarity, but businesses shouldn\u2019t interpret the current uncertainty as a green light to ignore copyright infractions. For compliance professionals, the unsettled nature of AI copyright law means companies<\/a> focus on the risks they can reasonably manage today. In my view, businesses should understand where their training data comes from, whether copyrighted material is subject to licenses or other restrictions, and maintain documentation showing how datasets are selected and used. Companies<\/a> should also consider testing AI systems for outputs that reproduce substantial portions of copyrighted works. While this measure cannot eliminate copyright risk, they can help companies identify potential issues before they develop in litigation.<\/p>\n

Greater legal clarity would nevertheless benefit both developers and copyright owners. The Copyright Office<\/a> has acknowledged that training AI entails balancing competing interests between technological progress and the protection of creators\u2019 rights. Guidance should be developed further by policymakers, rather than the years of case-by-case litigation to establish the boundaries of lawful acquisition and use of copyrighted material for the purpose of AI training. Until those boundaries are established, companies should treat copyright compliance as an element of AI development itself, not a problem to be solved only after the copyright owner sues.<\/p>\n

\n

\n","protected":false},"excerpt":{"rendered":"

Generative artificial intelligence (AI) has quickly become part of everyday business practices, but the technology presents an unresolved question for copyright law: can companies use copyrighted material to train artificial intelligence models without acquiring permission from copyright owners? That question is now central to major litigation involving OpenAI, Microsoft, The New York Times, authors, and other new organizations. As courts determine how traditional copyright principles apply to AI training, companies face a compliance challenge: create innovative AI systems while mitigating the risk of copyright infringement.<\/p>\n","protected":false},"author":183,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[4],"tags":[1303],"class_list":["post-6986","post","type-post","status-publish","format-standard","hentry","category-artificial-intelligence","tag-media"],"_links":{"self":[{"href":"https:\/\/blogs.luc.edu\/compliance\/index.php?rest_route=\/wp\/v2\/posts\/6986","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/blogs.luc.edu\/compliance\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blogs.luc.edu\/compliance\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blogs.luc.edu\/compliance\/index.php?rest_route=\/wp\/v2\/users\/183"}],"replies":[{"embeddable":true,"href":"https:\/\/blogs.luc.edu\/compliance\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=6986"}],"version-history":[{"count":1,"href":"https:\/\/blogs.luc.edu\/compliance\/index.php?rest_route=\/wp\/v2\/posts\/6986\/revisions"}],"predecessor-version":[{"id":6987,"href":"https:\/\/blogs.luc.edu\/compliance\/index.php?rest_route=\/wp\/v2\/posts\/6986\/revisions\/6987"}],"wp:attachment":[{"href":"https:\/\/blogs.luc.edu\/compliance\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=6986"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blogs.luc.edu\/compliance\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=6986"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blogs.luc.edu\/compliance\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=6986"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}