Global Artificial Intelligence Patent Survey

Article co-written by Yuri Levin-Schwartz, Ph.D., a law clerk at MBHB.

Research and development conducted worldwide is currently driving the so-called “Fourth Industrial Revolution” (referred to herein as “4IR”).[1] 4IR encompasses three major fields: (1) physical technologies, including autonomous vehicles, robotics, and 3D printing; (2) biological technologies, including genomic diagnostics, treatment, and engineering; and (3) digital technologies, including the Internet of Things (IoT) and blockchain.[2]

Artificial intelligence (AI), once centered primarily within the digital realm, has now been applied to all three major fields of 4IR, and beyond.[3] For example, AI-based medical devices are now able to automatically diagnose patients with various conditions, such as diabetic retinopathy or neurovascular abnormalities, based on image recognition analyses of routine diagnostic scans.[4] Furthermore, recent innovations in natural language processing have lowered the communication barrier between human and computational machines, enabling a variety of new AI-based technologies.[5]

As inventors have sought worldwide patent protection for their AI-based ideas, the number of international patent filings has expanded rapidly over the last few years.[6] Research indicates that most AI-related patent activity takes place in the United States and Asia. In fact, according to a 2016 study, approximately 75% of all AI-related patent publications in the world come from three jurisdictions: China, Japan, and the United States.[7] Although a majority of AI-related patents are being filed in these three countries, Europe and Australia are also seeing substantial increases in such patent filings.

This article explores AI-related patenting trends in various international jurisdictions and provides information on recent developments, common patentability issues, and tips for navigating the same. The article also suggests how such trends may impact the prosecution of AI-related patent applications in the United States.


China has overtaken the United States in the number of annual patent applications filed in the area of artificial intelligence.[8] Recent studies revealed that China accounts for around 37% of published AI-related patent applications[9] and about 22% of granted AI-related patents in the world.[10] Two main factors appear to be driving the number of Chinese patent applications. First, Chinese universities appear to be heavily engaged in AI-based research and innovation. China was the only jurisdiction where the universities filed more AI-related patent applications than its corporations.[11] Second, the amount of investment in Chinese AI-based companies is high and expectations for returns are even higher. Investors gave $4.5 billion to more than 200 Chinese AI-based companies between 2012 and 2017, and the country’s goal is to foster a $1 trillion AI industry by 2030.[12]

Subject matter eligibility requirements can be a barrier to prosecuting AI-related patent applications in the Chinese Patent Office.[13] Chinese patent law defines an invention as any new technical solution proposed for a product, a process, or the improvement thereof.[14] However, Chinese patent law prohibits patents directed towards “rules and methods for intellectual activities” and “methods for the diagnosis or treatment of diseases.”[15]

To protect ideas relating to intellectual activities in Chinese patent applications, practitioners suggest providing as many comprehensive details as possible regarding the invention in the disclosure. For example, Chinese patent applications should fully describe how the invention provides a stated functionality and how it achieves a desired technical effect.[16] Furthermore, regarding AI-assisted medical diagnosis and treatment methods, Chinese practitioners suggest focusing on the specific device utilized for diagnosis or treatment of diseases and/or the specific image or data processing method, as these aspects are more clearly established as patentable subject matter in Chinese patent law.[17] Furthermore, Chinese practitioners may utilize Swiss-type claims to protect the manufacture of a drug for a particular condition, e.g., “the use of substance X in the manufacture of a medicament for the treatment of condition Y.”[18]


The Japanese Patent Office (JPO) appears to be a patent-friendly forum for obtaining protection for AI-related innovations and consequently advantageous for patent owners. In particular, the JPO has taken steps to provide clear guidelines and assist applicants with claiming AI-related subject matter. The JPO considers inventions relating to AI and the IoT as business-related inventions.[19] Recent allowance rates for such business-related inventions were around 70%, which is almost as high as the allowance rate for patent applications in all other technological fields.[20]

The JPO’s allowance rate is significantly higher than the allowance rate for business method patents in the USPTO (12.7%)[21] and the European Patent Office (EPO).[22] This high allowance rate may be, at least in part, due to the JPO publishing case studies that cover claims relating to AI, IoT, and 3D printing in order to provide clear guidelines for the JPO’s examination procedures.[23]

Furthermore, in 2016, the JPO hosted a seminar entitled “How to Obtain Patent[s] Regarding Business-related Inventions.”[24] This seminar provided insight for drafting and prosecuting patent applications related to IoT and AI technology.[25] During the seminar, it was suggested that the higher allowance rate may be due to the consideration of both technical and non-technical subject matter during examination.[26] By contrast, non-technical features of an application are typically ignored by other examining authorities, such as the USPTO and the EPO.


In the Southeast Asia region, Singapore is the clear leader in AI-related patent filings, accounting for about 73% of that region’s AI-based patent publications.[27] Popular AI-related subject matter in Singapore includes innovations related to semiconductors, surgical tools, and digital data processing, and image and speech recognition.[28] Singapore’s subject matter eligibility guidelines for AI-related inventions are similar to that of China and the U.S. in that “[m]ethods that are considered mental acts or schemes are generally not inventions.”[29] Furthermore, similar to the EPO, examiners must determine whether there is an inventive step in the pending claims.[30] Some examples of AI-related patents allowed by the Singapore Intellectual Property Office include an invention for automatic aggregation of financial data and an invention that produces automatically cultured biological cells.[31]


While Europe may not be a leader in terms of number of AI-related patent filings, patent applications in Europe that relate to IoT and 4IR technologies grew at a 54% annualized rate from 2014 to 2017.[32] Germany, France, and Great Britain, in particular, have significantly increased the number of AI-related patent filings in recent years.[33]

The EPO has expressed a dedication to developing examination practices which are friendlier to computer implemented inventions (CII), which include AI-related subject matter. For example, recently the EPO adopted a new approach to interdisciplinary software patent applications. Specifically, such applications can now be examined by a team of three examiners with diverse technical backgrounds. Additionally, recognizing the urgency and fast-pace of the industry, the EPO has indicated it aims to speed up examination according to a 12-18 month timeline from filing to either allowance or final rejection.

Furthermore, in May 2018, the European Patent Office (EPO) held a conference entitled “Patenting Artificial Intelligence.”[34] The conference centered on the challenges and opportunities of patenting AI-related innovations. Speakers shared lessons learned and strategies for approaching AI related patents.

European examination of all computer related inventions includes a two-prong approach. First, examination includes an eligibility determination of whether the invention is directed towards, for example, a mathematical process or an abstract concept. Second, the invention must be directed towards a technical solution to a technical problem.[35] While the first prong traditionally represents a relatively low bar,[36] European practitioners can encounter more issues with regard to the second prong. During the EPO conference, a heavy emphasis was placed on claiming as specifically as possible. In particular, AI-related innovations should be described and claimed as being developed for a specific implementation. Furthermore, the AI-related idea should be shown to be motivated by technical considerations of the internal functioning of a computer, such as speed and/or computation load.[37]


While AI-related patent and published application numbers are not as statistically significant in Australia as compared to other jurisdictions, a recent court decision there may make Australia friendlier for AI-related patents. Recently, the Full Federal Court of Australia held that a patent application directed towards conversion of personalized information into one or more questions is directed “to a scheme or a business method that is not properly the subject of letters patent.” [38] However, in its opinion, the court suggests that the patent application may have been proper had the invention functioned “in the nature of an adviser or an artificial intelligence.”[39] Accordingly, such legal precedent may indicate broader patentability of AI-related inventions in Australia in the future.

United States

Patent applicants seeking to protect AI-related innovations in the United States face many of the same issues as those described in other jurisdictions. For instance, US practitioners certainly need to deal with the potential ineligibility of subject matter relating to abstract ideas and mental processes. As such, claiming AI-related innovations can be difficult in the US because, by some definitions, AI is arguably a computer-implemented mental process.[40] A common best practice suggested by practitioners for overcoming this problem is to claim narrowly and to keep the solution and the problem rooted in the technology and not solely in an algorithm or abstract method.[41]

Last year, the Northern District of California confirmed this approach in PurePredictive, Inc. v. H2O.AI, Inc.[42] The patent-at-issue involved automatically generating an “ensemble” of machine learning models. The court invalidated the patent claim under § 101 stating that it was directed towards “mathematical processes that not only could be performed by humans but also go to the general abstract concept of predictive analytics rather than any specific application.”[43]


As AI-based innovations become an ever-increasing presence in our daily lives, it seems inevitable that an expanding number of patent applications will be filed to protect such inventions. Certain international jurisdictions have been amenable to clarification and adjustment of their patent laws and examination procedures with regard to AI-related inventions, while others have been relatively unchanged. For example, while the subject matter eligibility of AI-related innovations is an open and evolving question in most major patent jurisdictions, it is promising to see both the Japanese and European patent offices working to clarify applicant expectations and streamline examination procedures in these areas. As US practitioners, it can only be hoped that the USPTO will establish similar initiatives to expedite and clarify AI-based patent prosecution as the Fourth Industrial Revolution marches forward.

© 2018 McDonnell Boehnen Hulbert & Berghoff LLP

snippets is a trademark of McDonnell Boehnen Hulbert & Berghoff LLP. All rights reserved. The information contained in this newsletter reflects the understanding and opinions of the author(s) and is provided to you for informational purposes only. It is not intended to and does not represent legal advice. MBHB LLP does not intend to create an attorney–client relationship by providing this information to you. The information in this publication is not a substitute for obtaining legal advice from an attorney licensed in your particular state. snippets may be considered attorney advertising in some states.

[1] Bernard Marr, The 4th Industrial Revolution is Here – Are You Ready?, Forbes (Aug. 13, 2018),

[2] Clare Dillon, AI: Our Changing World, EPO Patenting Artificial Intelligence Conference (May 30, 2018),$FILE/

[3] Daniel Araya and Creig Lamb, Surfing the 4th Industrial Revolution: Artificial intelligence and the liberal arts, The Brookings Institute (Apr. 11, 2017),

[4] Aaron Gin and Bryan Helwig, FDA Signals Fast-Track Approval for AI-Based Medical Devices, Bloomberg Law (May 9, 2018),

[5] Xinwen Zhang, The Evolution Of Natural Language Processing And Its Impact On AI, Forbes (Nov. 6, 2018),

[6] Michael Webb et al., Some Facts of High-Tech Patenting 1, (Nat’l Bureau of Econ. Research, Working Paper No. 24793, 2018).

[7] China Academy of Information and Communications Technology, 2018 World AI Industry Development Blue Book 13 (2018), available at

[8] Id.

[9] Id.

[10] China’s AI Patents Account for 22 Pct of World’s Total: Official, Xinhuanet (May 10, 2018),

[11] China’s Research Institutes File More AI Patents Than Businesses: Report, China IP Magazine (Aug. 14, 2018),

[12] Vikram Barhat, China is Determined to Steal A.I. Crown From US and Nothing, Not Even a Trade War, Will Stop It, CNBC (May 4, 2018),

[13] Yue Li and Xiaoming Zhang, Intellectual Property Protection for AI in China, Liu Shen & Associates (July 4, 2018),

[14] Patent Law of the People’s Republic of China, Chapter I, art. 2.

[15] Patent Law of the People’s Republic of China, Chapter II, art. 25.

[16] Yue Li and Xiaoming Zhang, supra note 13.

[17] Id.

[18] Yu Guo, China v US: What Can Be Patented in the Life Sciences Field?, Life Sciences Intellectual Property Review (Sept. 8, 2016),

[19] Nobuyuki Taniguchi, Recent Developments on Software-related Patents in Japan, Nakamura & Partners (Aug. 18, 2017),

[20] Id.

[21] Business Methods Allowance Rate, USPTO,

[22] Taniguchi, supra note 19.

[23] Id.

[24] Id.

[25] Id.

[26] Id.

[27] Artificial Intelligence in Southeast Asia, Clarivate Analytics (2018),

[28] Id.

[29] Intellectual Prop. Office of Singapore, Examination Guidelines For Patent Applications at IPOS 260 (2017), available at

[30] Mizuki Hashiguchi, The Global Artificial Intelligence Revolution Challenges Patent Eligibility Laws, 13 J. Bus. & Tech. L. 1, 26 (2017).

[31] Id. at 27.

[32] Yann Ménière, et al., Patents and the Fourth Industrial Revolution 11, European Patent Office (Dec. 2017).

[33] Id. at 13.

[34] EPO Hosts First Conference on Patenting Artificial Intelligence, European Patent Office (May 30, 2018),

[35] Yann Ménière, AI Inventions and the Fourth Industrial Revolution, EPO Patenting Artificial Intelligence Conference, May 30, 2018,

[36] Id.

[37]Koen Lievens, Patenting Artificial Intelligence, EPO Patenting Artificial Intelligence Conference, May 30, 2018,

[38] Commissioner of Patents v. RPL Central Pty Ltd, [2015] FCAFC 177 (FCR) (Austl.).

[39] Id.

[40] Benjamin Bai, Challenges in Drafting Patent Applications for AI-related Inventions, EPO Patenting Artificial Intelligence Conference, May 30, 2018,

[41] Id.

[42] No. 17-CV-03049-WHO, 2017 WL 3721480 (N.D. Cal. Aug. 29, 2017), aff’d 741 F. App’x 802 (Fed. Cir. 2018).

[43] Id. at *5.