Copyright Considerations Related to Generative AI
[20 Dec 2024]
The use of AI tools can be a tremendous asset in performing a broad range of tasks; however, it is important to ensure that these tools are used responsibly. This page focuses on copyright considerations, but for a broader perspective on the use of AI tools, please see the draft Framework for Responsible and Ethical Use of AI at the University of Alberta, linked on this page.
To maximize the responsible and productive use of AI tools, copyright considerations are important to ensure that the interests of the rights holders in copyright-protected works are appropriately respected, and to minimize the risk of copyright infringement and related legal liability arising from such use.
Copyright considerations related to the use of generative AI tools may best be viewed in terms of the following three principles, drawn from the principles identified in the Framework document noted above: transparency, accountability, and human oversight.
Copyright law has been continually adapting to and accommodating technological changes, and AI tools are no exception. To the extent practical, existing legal provisions, perspectives and interpretations will be employed to address copyright issues arising from the use of AI tools, as statutory reform and binding legal precedent can be slow processes.
There are a number of aspects of generative AI that have copyright implications.
-
The first of these is the development of large language models on which AI tools can train, and the use of text and data mining to acquire content for the construction of those models. The extent to which fair dealing might apply to copyright-protected content reproduced through such text and data mining is currently unclear. In the absence of an agreement with the rights holder, transparency and accountability dictate that, to the extent practical, the sources of text and data ingested by the model be well documented.
-
The second aspect is the uploading of someone else’s copyright-protected content into a generative AI tool by a user of that tool. A key copyright consideration would likely be whether that content will remain local, or whether it will become part of a broader dataset. Accountability suggests that distribution of third-party content should not take place in this way without authorization from the rights holder.
-
The third aspect is how to perceive and interpret copyright ownership of the outputs of generative AI tools. Current law suggests that only a human being can be recognized as an author for copyright purposes, thus works that involve an AI tool as part of the creative process must demonstrate a significant amount of direct human oversight in their creation for those works to be protected by copyright. Until there is more clarity around legal requirements, to ensure transparency and accountability, the Copyright Office recommends that the specific prompts used and the editing of output from the tool in the creation of the final work be well documented, and that the role of any generative AI tool used in the creative process be disclosed.
- Finally, the fourth aspect is where to place the responsibility for copyright infringement when the output of a generative AI tool is found to be infringing. Ultimately, this is an accountability that will likely come down to examining the chain of human oversight in the process to determine where responsibility for the specific infringement should reasonably lie.
Perhaps it should not be surprising that the common denominator for copyright considerations in relation to the use of AI tools seems to be accountability.
*****
In late 2023, the Federal Government conducted a Consultation on Copyright in the Age of Generative Artificial Intelligence. The University of Alberta Copyright Office submitted a response to this consultation, which is available here.
If you require additional information about the copyright implications of your use of AI tools, please contact the Copyright Office at copyright@ualberta.ca.