USC Libraries endeavor to negotiate “researcher-friendly” licensing terms that preserve fair use rights, which are essential to ensuring that scholars can lawfully make use of the materials we have procured. In support of this mission, researchers should ensure lawful acquisition of copyrighted content, especially when copying of that content is considered necessary for the purpose of research.
Refer to USC’s central hub for secure enterprise AI tools and note that uploading copyrighted or licensed material may be acceptable for non-commercial research, private study, or teaching, but only when using ITS-supported AI tools that do not copy or use the copyrighted material as training data.
Using other people’s work (regardless of whether it’s full or a partial section) of intellectual property (IP) that you don’t own or have permission to use requires careful consideration.
The question of whether AI outputs can hold copyright remains unsettled. In 2025, AI itself cannot be deemed an author, and therefore human authorship is essential for copyright. How or how much human involvement is needed is still being debated in the U.S. legal system.
** Cases regarding outputs contrast sharply with training data cases that focus on whether using copyrighted works to train AI constitutes infringement. See the Database of AI Litigation (DAIL) or further reading below.
Training AI models by using copyright-protected inputs falls within what courts have determined to be fair use, and particularly so when that AI training is for nonprofit educational or research purposes. Still, before you train or query any AI tool, you should consult USC Specific Guidelines.
Most of the digital research materials you access through the USC Libraries aren’t owned by USC. Instead, they’re owned by commercial publishers, academic societies, and other content providers who create and distribute scholarly resources. Each year, USC signs license agreements and pays significant subscription fees to publishers, so that you can access journals, books, and premium databases and tools for research and course work. When using any third-party copyrighted content, always consider whether you can rely on Fair Use. Copyright law gives you certain rights, including fair use for research and education. Contract law and USC licenses may override such rights.
Copyright broadly says that reading by robots doesn't count. Infringement is for humans only; when computers do it, it's fair use. In general, what this means is that text and data mining rules permit researchers to conduct computational analysis on materials to which they have access. Fair use for temporary copies made in the mining phase do not make copies that "count" under U.S. law. However, contract law and fair use rights are separate sources of authority. USC Libraries' license agreements with content vendors explicitly prohibit the use of robots, spiders, crawlers or other automated downloading programs, algorithms, or devices to continuously search, scrape, extract, or index data or metadata. Most publishers include Terms & Conditions links on their websites that detail specific restrictions on AI, TDM, and similar technologies - often reserving all rights for these uses.
Be mindful that USC Libraries can detect excessive downloads and violations of these agreements may result in immediate termination of access. USC acceptable use policy also allows the university to monitor and inspect files and communications on USC-owned technology resources at any time.
When engaging copyrighted material with AI – both as training input and AI-generated output – researchers should note that terms of service can and often do limit fair use. The U.S. Copyright Office Draft Generative AI Training report notes that using legally obtained sources for generative AI training for non-commercial research purposes is likely a fair use, however, the legal discussion is still ongoing.
The U.S. Copyright Office Report on Copyright and Artificial Intelligence analyzes copyright law and policy issues raised by artificial intelligence (AI). It's three parts were a response to congressional inquiries and expressions of interest from stakeholders.
Generative Artificial Intelligence and Copyright Law prepared by the Congressional Research Service (CRS), provides important background and context to discussions of AI and copyright. By law, CRS works exclusively for Congress, providing timely, objective, and authoritative research and analysis to committees and members of both the House and Senate.