Do Foreign Inventors Get a Fair Shot at Patent Grant in Canada? I Unleashed AI on 100,000 Patent Applications to Find Out

This summer I worked at an intellectual property firm in downtown Vancouver, immersed in the world of patent prosecution. My days were spent drafting correspondence and amendments for patent applications destined for the Canadian Intellectual Property Office (CIPO) and the United States Patent and Trademark Office (USPTO).

Our clients included both Canadian inventors filing patents directly in Canada, as well as inventors from many other countries, including the United States, Europe, and Asia – they file patents internationally, entering Canada through the Patent Cooperation Treaty (PCT).¹ As I worked on these patent applications, a nagging question began to form in my mind: Does CIPO treat foreign inventors equally as Canadian inventors when granting patents?

Prior research from other jurisdictions painted a mixed picture. A 2014 study by Webster² found that patent offices in Europe and Japan actually favored domestic applicants, while a 2017 study by Yang and Sonmez³ found that the United States and China patent offices treated applicants equally (and in China, foreign applicants even had slightly better patent grant outcomes). But no one had ever done such research for Canada.

In September, I decided to launch an AI-powered research project to turn my curiosity into a data-driven reality. I downloaded decades of raw patent bibliographic data from CIPO’s online database: gigabytes of information containing details about patent filings, grant outcomes, and timelines, etc. I selected all patents filed in Canada between 2010 and 2018 as the sample data for my research, totaling approximately 100,000 patent applications. Why this specific timeframe? Two reasons. First, Canada went through major legislative changes in patent prosecution in 2019 (Patent Act amendments),⁴ ending my sample dataset in 2018 made sure that I could study the Canadian patent system in a stable era of CIPO practices. Second, patents in Canada typically take 4-5 years to reach a final decision, ending my sample dataset in 2018 could avoid censoring bias from including too many pending patent applications whose final outcomes are not yet known.

I used Python to merge and clean my sample dataset, piecing together several gigabyte-sized files downloaded from CIPO that recorded various aspects of each patent application. The raw data was messy but after tons of data wrangling, I finally had a clean dataset with all the information I needed. With the assistance of Claude AI to handle the heavy computational lifting, I developed a logistic regression model to predict “grant outcome”, specifically testing whether an inventor’s filing country of origin would have any influence on the probability of getting their patent granted in Canada. I also developed a linear regression model to predict “pendency” (how long it takes for a patent to be granted) based on an inventor’s filing country of origin. My ultimate goal was to use AI and the regression models I developed to crunch a massive set of CIPO patent bibliographic data (around 100,000 applications) and answer the fundamental question that had been nagging me all summer: Does CIPO treat foreign inventors equally as Canadian inventors when granting patents?

In addition to the inventor’s filing country of origin, my regression models also took into account several other important factors that could potentially influence patent grant outcomes and pendency: 1) the technical field of the invention (e.g., Mechanical Engineering patents versus Chemistry & Metallurgy patents); 2) whether the application was filed with the assistance of a patent agent or firm; 3) whether it was an initial filing or based on a prior patent application; and 4) the year the application was filed.

The results of my AI-powered research were fascinating, and somewhat surprising.

Fascinating Findings

My logistic regression analysis revealed that Canadian-origin patent applications actually had a lower probability of being granted than applications filed by inventors outside of Canada. Specifically, Canadian applicants had a 34% lower probability of getting their patents granted.

The results from my linear regression analysis indicated that Canadian-origin applications experienced nearly identical pendency durations as foreign-origin applications, only 2.85 months faster out of an average of 54-month pendency durations. That’s essentially no difference.

I isolated the specific effect of an inventor’s filing country of origin, by controlling for all other factors in my regression models, such as the technical field, patent agent representation, initial filing status, and the year of filing. And the results showed that my regression models had modest-to-strong prediction power (AUC-ROC of 0.63 for the logistic regression model and adjusted R² of 0.42 for the linear regression model), meaning the results of my regression analysis can be trusted.

The lower grant rates and similar pendency durations for Canadian-origin patent applications indicate that Canada does not discriminate against foreign inventors and complies with the National Treatment Principle under the Paris Convention (NTPU),⁵ which requires domestic and foreign applicants to be treated equally. Because Canada is a major patent jurisdiction with many filings originating from outside the country, this finding is significant as it affirms the perceived fairness and integrity of the Canadian patent system.

On the other hand, before we conclude that CIPO actually favors foreign inventors, we should consider legitimate explanations for the lower grant rate of Canadian-origin patent applications. The difference likely reflects variations in patent prosecution strategies rather than any discriminatory practices adopted by CIPO. My sample dataset shows that between 2010 and 2018, most (89.8%) Canadian-origin patent applications were filed directly and only in Canada rather than in multiple countries through the PCT. Conversely, all patent applications originating outside Canada entered the Canadian National Phase through the PCT route.

The patent applications filed through the PCT route most likely have already been vetted through international search reports, reviewed by patent office such as the USPTO, and undergone claim amendments to address prior art and improved patentability. They also often represent inventions that are more commercially significant and worth the cost of filing internationally. Therefore, these PCT patent applications are likely of higher quality by the time they reached CIPO. It may be no surprise that they have higher grant probabilities in Canada.

Additional Findings

Several factors other than the inventor’s filing country of origin also revealed compelling insights about how the Canadian patent system works:

• Agent Representation Truly Matters, A Lot: Applications filed with the assistance of a patent agent or firm were 3.4 times more likely to be granted than those filed by inventors themselves. This aligns with my experience: professional patent drafting significantly improves claim clarity, prior art citations, and compliance with CIPO’s formal requirements. On the other hand, agent representation increased pendency duration by about 15 months. Why? Because patent agents advocate strongly for their clients, submitting detailed responses to office actions and pursuing extensive claim amendments to maximize patent scope and protection. All of this takes more time with CIPO.
• Technical Complexity Impacts Grant Outcomes and Pendency: Different technical fields showed dramatically different grant probabilities and pendency durations. For instance, Mechanical Engineering type of patent applications (IPC Class F) had much higher grant probabilities and shorter pendency than Chemistry and Metallurgy (IPC Class C) and Electricity (IPC Class H), which in turn performed better than Physics (IPC Class G). This reflects the varying complexity of prior art searches and examination requirements across different technologies.
• CIPO Is Getting Faster But Stricter: Later years showed declining grant probabilities (26% lower per year) but shorter pendency durations (7.3 months faster per year). This suggests that CIPO has improved patent examination efficiency over time while applying more rigorous patentability standards.

Limitations and Future Research

While my regression models had modest-to-strong explanatory power and captured the most important patterns, some aspects of both grant outcomes and pendency remain unexplained. What’s missing are factors such as patent examiner characteristics and patent application quality metrics.

Unlike the USPTO, which publishes detailed information about patent examiners’ characteristics, CIPO discloses almost nothing about its patent examiners. I had no way to account for whether an application was assigned to a senior versus a junior examiner, or whether the examiner had an unusually heavy workload. I used the technical field of each patent application as a proxy (different technical fields have different examiner pools), but it’s an imperfect solution.

Additionally, I could not directly measure patent application quality metrics, such as claim complexity, specification length, or the number of cited prior art references. While CIPO publishes the full patent applications, its database does not aggregate this information. Manually analyzing 100,000 patent applications for these quality metrics would be impossible. I therefore used agent representation as a proxy for quality, assuming that professionally drafted applications are generally of higher quality, though this remains only a partial solution.

Future research could address these limitations by somehow obtaining data on CIPO patent examiner-specific characteristics and using AI to scrape detailed patent applications in order to develop quality metrics.

Key Takeaways

The bottom line? Canada’s patent system appears to comply with the National Treatment Principle. There is no evidence of systematic bias favoring Canadian patent applicants. If anything, the bias may run slightly in the opposite direction. Foreign inventors can be confident that they’re getting a fair shake in Canada.

For Canadian inventors and businesses, the key takeaway is clear: invest in professional representation by hiring a patent agent or firm. The 3.4x increase in grant probability more than justifies the cost. Yes, it might add 15 months to pendency durations, but you’ll end up with a granted patent and isn’t that the whole point? Also, file internationally via the PCT route in multiple countries rather than only in Canada, since the higher likelihood of grant may justify the additional cost.

Final thoughts? Embrace AI! The power of AI and data science to support legal research is enormous. With Python, Claude AI, and massive data available in publicly accessible online databases, I was able to conduct an amazingly complicated legal analysis all by myself that would not have been possible just a few years ago. As we move forward into an increasingly AI-driven era, AI tools may become indispensable for empirical legal research and day-to-day legal practice.

AI Disclosure: AI-assisted tools were used for proofreading and stylistic editing, and statistical computations.

1. World Intellectual Property Organization, “PCT – The International Patent System”, online: WIPO https://www.wipo.int/en/web/pct-system.
2. Elizabeth Webster, Paul H Jensen & Alfons Palangkaraya, “Patent Examination Outcomes and the National Treatment Principle” (2014) 45:2 RAND J Econ 449 at 18.
3. Deli Yang & Mahmut (Maho) Sonmez, “Global norm of national treatment for patent uncertainties: A longitudinal comparison between the U.S. and China” (2017) 48:4 J World Bus 527.
4. Canadian Intellectual Property Office, “Changes to Patent Rules on October 30, 2019”, online: Government of Canada https://ised-isde.canada.ca/site/canadian-intellectual-property-office/en/patents/changes-patent-rules-october-30-2019.
5. Gaétan de Rassenfosse, “National Treatment: Principle or Practice?”, (4 November 2024), online: The Patentist https://www.thepatentist.com/p/national-treatment-principle-or-practice.