A Beginner’s Guide to Patenting Software and AI

Madelaine Lynch is a 3L JD Candidate at Osgoode Hall Law School, enrolled in Professor David Vaver’s 2021-2022 Intellectual Property Law & Technology Intensive Program. As part of the course requirements, students were asked to write a blog on a topic of their choice.

This article was written in affiliation with Madelaine’s placement at ventureLAB and published December 17, 2021, as part of ventureLAB’s IP Deep Dive Series.

This series provides an in-depth look at different types of intellectual property. As a founder, once you’ve determined what types of Intellectual Property (IP) you need to protect, it’s important to understand the details and requirements of each. A concrete understanding of each type of IP not only allows you to recognize what you may be able to protect, but will also help you to take full advantage of the right. 

‍This article is the second in a 5-part series. ventureLAB’s experts cover everything you need to know when filing for IP protection. Over the next six weeks, we’ll release articles on the following topics. Sign up for our newsletter to be the first to know when we release an article. 

  • Part 1: The Four Pillars of Patentability
  • Part 2: A Beginner’s Guide to Patenting Software and Artificial Intelligence
  • Part 3: Use It or Lose It: How to Acquire and Protect your Trademarks
  • Part 4: Don’t You Forget About Me: Protecting Your Company’s Copyrights 
  • Part 5: The IP Hidden Gems: Trade Secrets and Industrial Designs

Patenting software, and inventions related to Artificial Intelligence (AI) and machine learning, known as computer-implemented inventions (CII) in patent lingo, is a complicated and evolving area. The difficulty lies in overcoming Patent Office objections related to subject matter eligibility, which prohibits the patenting of abstract concepts and mathematical formulae (for more on this, see The Four Pillars of Patentability). 

CII, such as algorithms, are frequently deemed an abstract idea, and thus not suitable for patent protection. In order to overcome this obstacle the patent application must explain how the algorithm and the computer are interconnected such that together they form a single invention. To do this, the computer must be deemed an essential element of the invention.

This has been a historically tumultuous area of patent law. The 1981 Canadian Schlumberger case was the first to set out the rule that the use of a computer merely to automate calculations that could otherwise be done by hand is not patentable. This is because the inventive aspect is the calculation itself, not the combination of the calculation and computer. In 2011, Amazon was successful in obtaining a patent on their ‘one-click’ method. This was a controversial decision since not only was it a CII, but it was also considered to be a “business method”, which is notoriously difficult to patent in Canada.

Most recently, in 2020 the Canadian Federal Court decided the case Yves Choueifaty v Canada (Attorney General), which initiated a huge change in the way the patent office evaluates the essential elements that make up an invention. This resulted in the Canadian Intellectual Property Office releasing new guidelines on how patent examiners should approach patentable subject matter, particularly for CII. 

Before this decision, patenting was particularly difficult for CII because patent examiners would frequently determine the computer itself was not an essential element of the invention. As a result, what remained was an abstract idea (i.e. the software or algorithm itself) which had no physical substance and was thus deemed unpatentable. 

Now, all elements of an invention are presumed to be essential. Thus, if you include a computer in the claims, it will be assessed as if that computer is essential to the functioning of the invention. This is great news for inventors, since it reduces the likelihood that your CII will be automatically deemed unpatentable. 

Unfortunately, the challenges do not stop here. Having a computer viewed as an essential element of the invention is only the first step in patenting software. Next the patent examiner will look at whether the computer works together with the other essential elements of the invention. Together, the computer and the other elements must form an actual invention that has physicality. This physicality can be physical existence, or the invention can produce a physical effect or change.


The mere use of a computer in a well-known manner, such as to execute the code, will not be sufficient to render the invention patentable. This is because any computer could do that job, and the computer is simply a means to execute the algorithm, which on its own is an abstract theorem or disembodied idea. However, if in processing the algorithm the function of the computer itself is improved, this could be considered a physical effect and thus a unified invention.

In the US the test for patentability is different, but ultimately the outcome is similar. You cannot obtain a patent on a purely abstract idea and merely using a computer in a generic way will not be enough to overcome this hurdle. 

In Europe, the patent office takes a slightly different approach. It is relatively easy to overcome the subject matter barrier as the “any hardware” approach means that simply including a computer in the claim is a first step towards satisfying patentability. However, the patent office also requires a technological solution to a technological problem, which looks only at the technical features that contribute to the solution. Unlike Canada which now presumes most, if not all, elements are essential, a non-technical feature which doesn’t interact with the technical subject matter is essentially ignored. 

Patenting software is a very challenging field. The summary above is a very basic introduction to the field. It is best to seek the advice of a patent professional who can advise on the specific needs of your invention and adapt it for each jurisdiction. A patent professional will consider how your invention integrates the use of a computer, and how the computer and algorithm cooperate to produce a physical effect.

One Comment
  1. Great article Maddie. While we wait for some more cases to help define the FC’s decision in Yves Choueifaty v Canada (Attorney General), CIPO has issued guidance on this matter (https://www.ic.gc.ca/eic/site/cipointernet-internetopic.nsf/eng/wr04860.html). The examples provided show and state that CIIs produce a physical change and effect and are therefore patentable if they “us[e] less processing power (fewer instructions) than known methods.” How that argument applies in prosecution is still being worked out.

Leave a reply

Your email address will not be published. Required fields are marked *

10 + 6 =