About
Overview Endorsements
Service
Media Center
Insights Events Recognitions
People
Publications
Publications Laws & Regulations Perspectives Newsletter
Contact Us
CNIPA updates patent examination guidelines to address emerging technology, regulate dual filing and invalidations
Time:Nov 27 2025

Xiaoling Duan, 26 November 2025, first published by IAM

 

On 13 November 2025, the China National IP Administration (CNIPA) enacted the amendment to the Patent Examination Guidelines, which will become effective on 1 January 2026. The amendment aims to bring the patent examination practice up to date with emerging technology and business forms, covering, in particular, the legality and ethics of AI and Big Data applications, sufficient disclosure and inventiveness assessment of algorithm/data-related inventions, streaming media for video-coding BitTorrent, as well as subject matter of plant variety and biological breeding patents.

 

Inventions involving AI, algorithm and data

 

Subject matter

 

The newly added text in section 6(1) of Chapter 9(2) stipulates that invention patent applications containing algorithm or business method features should not be granted patent rights, provided that the data collection, tag management, rule setting or recommended decision making contains any content that violates laws or social ethics or harms public interest.

 

The CNIPA has included an example of data collected through facial recognition being used for precise marketing of commodities in business premises, such as shopping malls. That application would not be allowed. The rationale behind this is that China’s Personal Information Protection Law prescribes that devices for image capture and personal identification will be installed in public areas with prominent warning signs, if it is necessary for maintaining public safety – and the collected images and identification information is to be used for this sole purpose. Where the aforesaid collected data is to be used for other purposes, individual consent must be obtained.

 

Another example of an ineligible application relates to an emergency decision-making model for autonomous vehicles. By using pedestrians’ gender and age as analytic data, the trained model is used to determine which objects should be protected in a scenario of unavoidable obstacles. This application would be rejected because, according to the CNIPA, the selection of protected and collided objects based on gender and age goes against the public's equal right to life. In addition, this decision-making approach will reinforce gender and age biases in society, raise public concerns about the safety of public transportation and undermine public trust in technology and social order.

 

Inventiveness

 

The newly added text in section 6(2) of Chapter 9(2) explains the criteria for assessing inventiveness of applications involving algorithm or business methods. It is crucial that algorithmic and technical features support and interact with each other in terms of functionality for the contribution to inventive step. If an invention only involves the simple replacement of some elements – like collecting data from objects completely different from the prior art, without making adjustments or improvements to deep learning, model construction or training processes – it will be deemed non-inventive.

 

Sufficient disclosure

 

The newly revised section 6(3) of Chapter 9(2) details the requirements for drafting a patent description in relation to technical solutions for building or training AI models. The CNIPA mandates that the description clearly depicts the modules, hierarchies and connection relationships essential for the models, as well as the specific steps and parameters required for training. Where the invention involves the application of an AI model or algorithm in a specific field or scenario, it is generally necessary to clearly describe how the model or algorithm is combined with such field or scenario and how its input and output data are configured to indicate its inherent correlation, so that a person skilled in the art can implement the invention’s solution in accordance with the disclosed contents of the description.

 

Inventions involving BitTorrent

 

The newly added section 7 of Chapter 9(2) outlines the examination rules governing inventions involving BitTorrent. The CNIPA reiterates that BitTorrent, if being claimed as a standalone subject matter, shall fall under unpatentable rules and methods of intellectual activities. For inventions involving BitTorrent, the CNIPA illustrates how they can qualify as eligible subject matter. In the field of digital video-coding and decoding technology, if a specific method for generating BitTorrent is listed as the technical solution, and the method for storing or transmitting and the computer-readable storage medium for storing the said BitTorrent can achieve optimised configuration of storage or transmission resources, then such storage or transmission method and computer-readable storage medium are patentable subject matter.

 

Subject matter of plant variety and biological breeding patents

 

The newly amended section 9 of Chapter 10(2) broadens the scope of eligible subject matter in plant varieties and thus enhances patent protection in this industry.

 

A plant is defined as a typically non-moving organism capable of utilising photosynthesis to synthesise carbohydrates and proteins from inorganic substances such as water, carbon dioxide and inorganic salts to sustain its survival. Untreated, naturally occurring wild plants found by humans from the natural world are excluded from patent protection. When the wild plants are artificially selected or improved and have value in industrial applications, the plant per se is no longer part of the scientific discovery.

 

In the newly amended section 4(4), Chapter 1(2) “plant varieties” is redefined as a plant population artificially selected or discovered and improved, with consistent morphological and biological characteristics and relatively stable genetic traits. Plants and their reproductive materials obtained by artificial breeding or improvement from the discovered wild plants shall not be deemed as plant varieties, in case of absence of consistent morphological and biological characteristics or relatively stable genetic traits in their population.

 

A genetically modified plant produced through recombinant DNA technology of genetic engineering – if it still falls under the category of "plant varieties" – remains ineligible subject matter. New plant varieties can be protected through the Regulations on the Protection of New Plant Varieties in China.

 

Assessment of inventiveness in general

 

The amended section 6(4) of Chapter 4(2) emphasises the existing inventiveness assessment principle – the claimed technical solution should be subject to the inventiveness assessment test in its entirety, rather than a test of inventiveness of certain technical features. The CNIPA also explicitly states that in principle, the inclusion of features not contributing to the solution of technical problems shall have no bearing on the inventiveness assessment. The revision is expected to dissuade the practice of piling excessive features in patent applications.

 

Dual filing of inventions and utility model patents

 

The CNIPA used to allow the same applicant to file a utility model and an invention patent application in parallel for the same invention on the same day to secure patent rights with varying scopes of protection. This practice will not be allowed when the revised guidelines enter into force, as the amended section 6(2)(2) of Chapter 3(2) makes it clear that only one patent can be granted for dual filings, and once an invention patent application is granted, the applicant must abandon the utility model. In the filing process, the applicant is obligated to declare the concurrent invention and utility model patent applications.

 

Invalidation procedure

 

The amended guidelines introduce a notable change to the invalidation procedure. Section 4(1)(6) of Chapter 1(1) disqualifies any patent agencies or attorneys from acting as an invalidation petitioner. Offenders will be subject to fines, temporary suspension of new patent businesses and even licence revocation.

 

The amended section 3(2) of Chapter 3(4) also stipulates that an invalidation request that fails to reflect the petitioner’s true intentions shall be dismissed. The new rule seems to be targeting illegal or malicious patent invalidation actions by limiting the use of a ‘straw man’ in initiating an invalidation request. However, it remains to be seen how the new rule will be implemented in practice.

 

It is also worth noting that section 2(1) of Chapter 1(1) cranks up efforts to apply the non bis in idem principle in curtailing repeated invalidation actions based on grounds and/or evidence not only identical but also substantially identical with those submitted in prior invalidation proceedings, in which the CNIPA has already made an invalidation decision.

 

Eligible inventors

 

The amended guidelines also address eligibility surrounding inventors. When filing patent applications, applicants are required to fill in the authentic inventors and the identity information of all inventors. The CNIPA further mandates that an inventor must be a natural person, explicitly banning listing any entity, group or AI independently or therein as inventors. Patent agencies are tasked to verify the applicant's identity information and contact information in the request form.