China set to introduce further protection for AI-related innovations

Managing IP is part of Legal Benchmarking Limited, 1-2 Paris Gardens, London, SE1 8ND

Copyright © Legal Benchmarking Limited and its affiliated companies 2026

Accessibility | Terms of Use | Privacy Policy | Modern Slavery Statement

China set to introduce further protection for AI-related innovations

Sponsored by

Liu Shen logo.png
2722500c-9e96-43d5-974f-256703ab6678yurii-stupen-3tgnwvc7cjk-unsplash-1-min.jpg

Xiaoming Zhang of Liu Shen & Associates takes a closer look at the Draft for Comments to the Patent Examination Guidelines and discusses its positive approach towards AI-related applications

On November 10 2020, the China National Intellectual Property Administration (CNIPA) issued a revision to the Patent Examination Guidelines (Draft for Comments). The revision may enter into force alongside the revised Patent Law and the Implementing Regulations of the Patent Law on June 1 2021.



As discussed in a previous article, the Patent Examination Guidelines had been updated in February 2020, and aimed to address prominent issues in the artificial intelligence (AI) field including eligibility, novelty/inventiveness, and sufficient disclosure. It is generally uncommon for the CNIPA to release updates twice within a short period of time. Compared with the last update, the key differences introduced by the Draft for Comments are outlined below.

New type of subject matter: ‘Computer program’ product

For a long time in the past, a computer-implemented invention could only be protected under its subject matter such as method, apparatus, device or system, but a ‘computer program’ per se was not a patentable subject matter before the CNIPA. In the amendments of 2017, the subject matter of a computer-readable medium was finally permitted. A ‘computer program’ product, interpreted as a software product to solve problems via a computer program, has been introduced for the first time.



The draft provides four kinds of subject matters for the AI-related computer-implemented application, as follows:

  • A method for removing image noise, comprising: (steps…);

  • A computer device/apparatus/system comprising a memory, a processor and computer program stored on the memory, characterised in that the processor executes the computer program to fulfil the method as claimed by Claim 1;

  • A computer-readable storage medium on which computer program/instructions are stored, characterised in that when executed by a processor, the computer program/instructions cause the processor to fulfil the method as claimed by Claim 1; and

  • A computer program comprising computer program/instructions, characterised in that when executed by a processor, the computer program/instructions cause the processor to fulfil the method as claimed by Claim 1.


Specific examination rule regarding eligibility

As for AI-related computer-implemented applications, the eligibility judgment is always the key issue to be solved. In accordance with the former version of Patent Examination Guidelines, a solution shall adopt a technical means to solve a technical problem and achieve a technical effect, to render the solution to be a technical solution, as prescribed by Article 2.2 of the Patent Law of China.



In the past, it was stated that the technical means should be performed by a physical element, hardware or device, while the steps of an AI-related algorithm performed by a computer executing the computer instructions were commonly not considered as the qualified technical means, meaning that AI-related algorithm claims were ineligible.



In the Draft for Comments, “a step/means fulfilled by a computer” is determined to be a qualified technical means for the first time. Meanwhile, it is interpreted that “a step/means fulfilled by a computer” comprises of two cases. One case is that the claim defines a hardware and the content controlled or processed by the hardware executing the computer program. The other case is that even though no hardware has been defined directly, the solution fulfilled by the computer program can reflect the content controlled or processed by the hardware. Thus, most of the AI-related algorithm applications will be eligible since they at least adopt the newly introduced “a step/means fulfilled by a computer”.

Algorithm feature’s contribution to inventiveness

The Draft for Comments clarifies that if an algorithm feature leads to an improvement of the internal performance of the computer, such as reducing data storage, reducing data transfer or speeding up hardware processing, the algorithm feature will be taken into consideration when evaluating inventiveness.



Similarly, if the improvement of the user experience is caused by an algorithm feature, the algorithm feature will play a role in the judgment of inventiveness.

The remaining challenges

From practical experience and an understanding of the examples given in the Draft for Comments, it can be stated that the business rules and methods involved in financial solutions are still precarious to apply for, and may still be subject to increased rigorous examination. Thus, the usage of specific financial terms such as ‘bitcoin’ and ‘virtual money’ in the application documents, remains better avoided for applicants.



With the implementation of the newly revised Patent Examination Guidelines, the hurdle of determining an eligible subject matter can be resolved more easily in China than in other jurisdictions. A solution with at least one technical means, even if it is “a step/means fulfilled by a computer”, will not be rejected as being ineligible. So far, the CNIPA will be the only authority that prescribes that algorithm features, business rules and user experience should be taken into consideration, along with the technical features as a whole, when evaluating the substantial contribution of an application.



In summary, the CNIPA has adopted a positive attitude towards AI-related applications, which may benefit applicants in the field.

Xiaoming Zhang

Partner, Liu Shen & Associates

E: xmzhang@liu-shen.com

more from across site and SHARED ros bottom lb

More from across our site

James Hill, general counsel at Norwich City FC, reveals how he balances fan engagement with brand enforcement, and when he calls on IP firms for advice
In the second of a two-part article, Gabrielle Faure-André and Stéphanie Garçon at Santarelli unpick EPO, UPC and French case law to assess the importance of clinical development timelines in inventive step analyses
Public figures are turning to trademark protection to combat the threat of AI deepfakes and are monetising their brand through licensing deals, a trend that law firms are keen to capitalise on
News of Avanci Video signing its first video licence and a win for patent innovators in Australia were also among the top talking points
Tom Melsheimer, part of a nine-partner team to join King & Spalding from Winston & Strawn, says the move reflects Texas’s appeal as a venue for high-stakes patent litigation
AI patents and dairy trademarks are at the centre of two judgments to be handed down next week
Jennifer Che explains how taking on the managing director role at her firm has offered a new perspective, and why Hong Kong is seeing a life sciences boom
AG Barr acquires drinks makers Fentimans and Frobishers, in deals worth more than £50m in total
Tarun Khurana at Khurana & Khurana says corporates must take the lead if patent filing activity is to truly translate into innovation
Michael Moore, head of legal at Glean Technologies, discusses how in-house IP teams can use AI while protecting enforceability
Gift this article