In this Realm, Copyright is King
IP comes in 4 broad categories, each offering its own angle of protection, and all of which can be tools for software and AI: Patents, Copyright, Industrial Designs and Trademarks. In software and AI research, copyright gets the spotlight. To briefly illustrate why: copyright is the basis for every Creative Commons, Open Source, and End-User licence agreement you click through (or require others to click through) to use a software or access code.
Copyright protects the fixed expression of an idea- the specific way it is written or drawn - rather than the idea itself. In the realm of AI and software, this may manifest in several ways:
- Software Code: often considered "weak" protection because if someone else writes different code to achieve the same functional result, they are generally not infringing.
- Structure and Organization: if the sequence or organization of any element of a software is sufficiently original, it may also fall under copyright protection.
- Data vs. Datasets: raw data itself is not copyrightable; it simply "exists". However, how you compile, present, or annotate that data can be protected as a creative expression.
Free of Charge or Free to Charge Forth
A common misconception in software development is that "free" means "no strings attached." In reality, "free (no charge)” does not mean "free (no obligations)". Similarly, paying for a service/existing code also does not necessarily mean that you’re free of obligations.
When using open-source code, Creative Commons (CC) assets, or even proprietary data, you are entering a contractual relationship on how the material can be used. Some licenses may be highly permissive (allowing any use), while others might be "copyleft" (requiring you to share your results under the same terms) or strictly non-commercial. And while there are types of research exemptions in IP, in copyright it may “lose” this status if there is an eventual intent to commercialize the project.
Quicklicence decode: “CC BY” = credit required, “NC” = commercial only, “ND” = no adaptations, “SA” = share adaptations under the same licence. Attribution alone may not satisfy the licence if other conditions apply.
Consider the example of using published works for training an AI model. If you cannot quickly confirm reuse rights, review the publisher’s permissions page. Many Canadian and international publishers route requests through automated services such asRightsLink or the Copyright Clearance Center, which may confirm whether your intended use is permitted or allow you to request a licence. In Canada, collective societies such as Access Copyright (outside Quebec) or Copibec (in Quebec) may also assist with certain permissions. If these routes do not clearly cover your intended use, you may need to contact the publisher directly. Terms and fees can vary depending on the type of use, and your technology transfer office is there to assist you with this.
To complicate things further, AI and software’s rapid innovation is currently challenging established legal frameworks, which leaves many key questions unanswered, leaving us to navigate a murky and shifting landscape..
Good Form for AI Researchers
The field is currently forging into uncharted territory making it difficult (if not impossible) to define the landscape. For example, a major question these days relates to scraping published materials for AI analysis. Unfortunately, even if restricted to academic purposes, this is one of the areas where the legal framework has not been clarified, and we are awaiting resolution of court cases whether in Canada or elsewhere, such as the ongoing dispute between OpenAI and the New York Times on the matter.
In the view of IP experts, each project is more nuanced than may meet the eye and deserves guided attention as early as possible. However, despite the murky depths of this realm, the map does have some boundary lines to help guide us.
- Define Your Goals: Without understanding your goals, you cannot see which paths are suitable to take. Are you aiming for a spin-out company, to provide an open-source tool, or do you strictly wish to publish academically ? Your end goal helps determine the ways IP may affect you.
- Beware of Terms of Use: Whether building off existing code, data, images, questionnaires, or other materials, whether publically or freelyavailableor if you’ve hired an external party to build it, question what you are allowed to do with it before committing to it. With software and AI, it is notoriously difficult to "go back in time" to fix IP issues once they are baked into a project.
- Document and Record: Keep meticulous records of where your code, data, and libraries came from and what licences they carry with annotated code and clear file names. Note collaborators, contributors and other human interventions. Clear documentation of human input may also help demonstrate authorship where copyright protection is important. With this, your AI and all possible obligations will be well supported to confidently pursue your goals, particularly if commercialization is in the plan.
- Mind the "Hidden Edge": This is particularly important if IP protection is warranted to help achieve your goal. You can often maintain academic integrity and publish your findings while keeping a small, specific piece quiet to allow for future patenting or protection.
- Consult the Experts: Navigating this IP side quest is a partnership with your technology transfer office. The burden of IP strategy, compliance and navigation is not yours to carry alone, and much relief can be provided by turning to resources embedded in the field.
This is general information, not legal advice. If you have questions about a specific project or are interested in learning more about IP strategy or how to navigate the path to real-world applications of your research, please reach out to your technology transfer office early and come back often.They are your partners on this quest!
This article was contributed by Leah Labib, Technology Transfer Officer with the Ottawa Hospital Research Institute, where she helps researchers protect and commercialize innovations through patents, licensing and research agreements.
