Copyright Resources

Generative AI tools like ChatGPT and Midjourney (to name just a couple) bring up a lot of questions in the academic environment, not least around copyright issues.  This technology is evolving rapidly, as is our understanding of its role in teaching, learning, and researching.  Generative AI is the subject of many active lawsuits and significant debate.  We will do our best to provide the most current information and guidance about copyright and AI on this guide.

No, not in the United States.  The current guidance from the US Copyright Office is very clear that works generated by AI tools are not subject copyright because only works created by humans are copyrightable.  In cases where you add expressive content of your own to AI generated content, the copyright in your contributions would be owned by you but the contributions generated by the AI would not be subject to copyright.

If you want to use material that was generated by AI whether that is text or an image or some other content:

• Make sure you are using the material ethically and in line with any classroom policies your instructor has given you.  This guide from UC San Diego gives you some excellent things to think about regarding AI and Academic Integrity.
• Make sure you've fact-checked any assertions the AI text is making and verified any sources that it references.  AI tools do not always share accurate information or viable sources.
• Cite the material properly according to the citation format that you are using for your project:
  • Citing AI in MLA format
  • Citing AI in APA format
  • Citing AI in Chicago style
  • If you are not using one of these standard citation formats, ask a librarian or your instructor for guidance.  Remember, the goal of citation is to communicate with clarity about the source of information so it is important to provide a citation even though copyright does not apply.

Does the training of generative AI tools using copyright material violate copyright?

Creating a useful generative AI tool requires that these tools are "trained" on massive pools of data.  All of the existing major AI tools were trained on data that includes massive amounts of copyrighted materials and this training is nearly always done without the permission of the copyright holders in that content.  So is this a violation of copyright?  This is currently the subject of intense debate and many open lawsuits from authors, artists, and other copyright holders who believe that such training violates their copyrights.  Others, including the makers of AI tools, feel that such training activities constitute a transformative fair use and cite such previous cases as  Authors Guild v. HathiTrust, Authors Guild v. Google, and A.V. ex rel. Vanderhye v. iParadigms where computer programs used text data mining to create new tools as relevant precedents.