The use of generative artificial intelligence (GenAI) services is common at the University of Windsor and across the higher education sector. In support of these developments, academic libraries seek to empower researchers to utilize AI tools in their work for all legal purposes, supporting the core missions of research, teaching, learning, and equitable access to information.
However, GenAI and related technologies are subject to existing laws and regulations in Canada, such as intellectual property, copyright and privacy laws. The legal status of various issues pertaining to GenAI is currently unsettled and continue to evolve.
In the context of academic libraries and the use of scholarly resources in teaching and research, one helpful way to organize how we think about copyright and GenAI is to distinguish between copyright issues related to inputs, the content we provide to AI tools, and those related to outputs, the content generated through our use of AI.
Inputs: Use of copyrighted works as input or training data
In Canada, there remains uncertainty regarding how familiar exceptions to obtaining permission, such as fair dealing, apply in this context. The government of Canada launched the Consultation on Copyright in the Age of Generative Artificial Intelligence to gather feedback from stakeholders. The results of this consultation made clear that there are a variety of perspectives and concerns related to GenAI that illustrate the tension between protecting rightsholders and creators and enabling training of AI models.In academic libraries, this raises concerns about using copyrighted materials that the library licenses (journal articles, books, videos, etc.) in the context of GenAI. Given this, the library is attempting to provide some guidance to users in this area. In general,
Can I use full-text library licensed resources in Generative AI tools like ChatGPT?
No. The library signs contracts with vendors and publishers, and the terms and conditions of these licenses outline what users are permitted to do when accessing content. These agreements do not allow for uploading copyright-protected works to third-party platforms, including generative AI tools.
Consult Leddy Library’s Acceptable Use of Online Resources policy for general information on how the library’s electronic resources may be used.
- Can I use full-text library licensed resources in closed or self-hosted environment?
Maybe. For some select resources the library has negotiated the right to use the licensed content with artificial intelligence tools for non-commercial, research and academic purposes. In these cases, the use is restricted to AI tools in a closed or self-hosted environment, where only other authorized users of the institution can access the content. Such use cannot train the algorithm of a third-party AI tool or result in any portions of this content being retained by the third-party tool (e.g., with a third-party AI tool like ChatGPT).
Please contact the Leddy Library for assistance in identifying which resources are available for such use, and the specific license terms. Consult Leddy Library’s Acceptable Use of Online Resources policy for general information on how the library’s electronic resources may be used.
Can I use full-text, openly licensed material (e.g. Creative Commons) and works in the public domain in Generative AI tools?
Material that is in the Public Domain may generally be used, however, you should confirm whether any license applies to the material.
Outputs: Copyright status of content created by GenAI
Copyright protection applies to all original work set out in the copyright act and as such creators of original works can benefit from copyright in their work. However it is not clear that using GenAI to generate content is subject to copyright protection.In Canada, existing copyright jurisprudence requires a human or group of humans as creators of that work. When using GenAI to generate an output it is not always clear that there is sufficient human input and originality in the process such that the output is subject to copyright. There are varying degrees of human input and originality in GenAI content generation and it can be complex process to determine whether copyright applies to a given output. As a result, users of GenAI products should exercise caution with the output of GenAI and any assumptions about the ownership and copyright of the resulting outputs.
Copyright infringement resulting from the use of GenAI is also possible. Image models are capable of generating very close duplicates of training images. Sharing, publishing or reproducing something as part of your work that was created using GenAI can potentially result in copyright infringement if the output is determined to be a substantial reproduction or of an existing work. Users should exercise caution when distributing content that utilizes outputs created with GenAI, particularly when prompting the model to mimic or copy a specific artistic style or literary voice.
Attribution of Sources
Giving credit to others' original work through attribution is fundamental to academic and scholarly work. GenAI models are trained on large amounts of text data from the internet, books, articles, and other sources. However, specific sources are not typically documented or easily traceable5 in a given output. This makes traditional scholarly attribution difficult or impossible.Thanks to Carleton University Library, University of Toronto Library, University of Manitoba and University of Waterloo's copyright office from which some of this content was adapted.
Updated: 2026-10-20
Connect with your library