Artificial Intelligence

Copyright and Artificial Intelligence – Part 2: Copyrightability, published by the Library of Congress, offers essential guidance for understanding how traditional copyright principles apply in an era of rapidly evolving generative technologies. This resource explains how U.S. copyright law evaluates authorship, originality, and protectable expression when AI systems contribute to the creation of a work. By clarifying what aspects of AI-assisted outputs may qualify for protection—and what remains outside the scope of copyright—it helps students, educators, and creators navigate the shifting legal landscape with greater confidence and integrity.

Generative AI (text, image, code, audio) is transforming scholarship, teaching, and creative practice—but it also raises new copyright and academic integrity questions. Because the law is still developing, students and faculty should follow existing copyright principles, university policies, and best practices when using AI tools. 

Authorship and Ownership 

• Human authorship required: U.S. law currently recognizes copyright only in works created by humans, not machines (U.S. Copyright Office). 
• AI-assisted works: Copyright applies only to the original human contribution (e.g., selection, editing, arrangement). AI-generated portions alone may not be protected (Congressional Research Service). 
• UMGC guidance: Clearly identify which parts of a project are human-authored and which are AI-assisted. Do not claim exclusive rights over wholly AI-generated material. 

Using Copyrighted Works as Input  

• Uploading copyrighted text, images, or datasets into AI tools may count as reproduction or derivative use. 
• Avoid submitting full works unless licensed or justified under fair use. 
• Always check the tool’s terms of service and use opt-out features when available (Columbia University AI Policy). 

Fair Use and AI  

• The four-factor test still applies: purpose, nature, amount, and effect (U.S. Copyright Office). 
• The legality of using copyrighted works to train AI is unsettled. Courts and the Copyright Office caution against overreliance on fair use when outputs closely resemble original works (Skadden, Mayer Brown). 
• Outputs that replicate training content or use “substantial” portions of copyrighted material are less likely to qualify as fair use. 

Transparency and Academic Integrity 

• Disclose AI use in assignments, publications, and research when it contributes substantially (e.g., text drafting, image creation). 
• Follow course-specific policies on whether AI tools are permitted; unauthorized use may count as academic dishonesty (University of Michigan AI Guidelines). 
• Keep records of prompts, tool versions, and edits for accountability. 

Licensing, Permissions, and Rights Clearance 

• Secure licenses or permissions if incorporating copyrighted content in AI outputs and fair use is uncertain. 
• Note that some publishers are adding “no AI training” clauses in their licenses (The Verge). 
• Prefer open-licensed, public domain, or original materials for training and outputs. 

Ethical Best Practices 

• Avoid asking AI to mimic or replicate distinctive copyrighted works. 
• Review outputs for accuracy, bias, or verbatim reuse of copyrighted material. 
• Use AI to support—not replace—scholarly rigor and creative integrity. 

When in Doubt 

If you are a UMGC faculty member and are unsure about copyright, licensing, or proper AI use in your work: 

• Consult UMGC Library Support who will work with the Integrative Learning Design Team and the Office of Legal Affairs 
• Document your process and decisions. 
• Err on the side of caution in public, commercial, or research contexts.