---

The rapid development of AI technology has injected new impetus into research and development (R&D) activities in high-tech industries. AI’s powerful advantages are making it an indispensable path in R&D of every technology. Given the increasing number of patent applications involving AI technology filed in China, the China National Intellectual Property Administration (CNIPA) officially released the Decision on Amending the Guidelines for Patent Examination¹ on November 13, 2025, with the aim of refining the examination standards for patent applications involving AI technology. This article will elaborate on key issues to note when filing patents applications involving AI technology in China in view of the latest AI-related examination standards stipulated by CNIPA's said Decision, as well as CNIPA’s recently published research articles in the relevant area.

 

For the purpose of this article, an invention involving AI technology refers to an invention where one or more stages of the process utilize AI technology, or where the entire process utilizes AI technology. In such a research process, there is both participation from researchers and the application of AI technology.

 

I. Inventorship

The Decision explicitly states that only natural persons can be listed as inventors for a patent or patent application, and an AI technology cannot be an inventor. This position is in consistent to those hold by other primary countries. This position also implies that the CNIPA believes that AI technology, including generative AI and other computational models, is merely an R&D tool, like other experimental equipment and computer software, utilized by humans, and its actions depend on the instructions and prompts input by the user. Even if AI technology has a high degree of involvement in R&D activities, the initiator of its actions remains a natural person.

 

II. Ethics and Public Interest

In order to comply with the stipulation in Article 5, Paragraph 1 of the Chinese Patent Law that: “No patent right shall be granted for any invention-creation that violates laws or social morality or that is detrimental to the public interests”, and to echo China’s national AI policy in moral aspect, that is, “Intelligence For Good”, the Decision specifically introduces some relevant rules to require that the ethic and public interest should be  examined on the entire application document, i.e., not only the claims, but also the content in the description.

 

In an example relating to a mattress sales system based on big data¹, the invention obtains customers’ genuine feelings about the mattresses and achieve precise marketing. For this purpose, the system installs camera devices in public places to film customers’ facial features and obtain customer identity information without their awareness. However, this data collection activity is not for public safety and does not obtain individual consent of the customers. Therefore, this invention may violate the relevant Chinese privacy law and should not be granted.

 

In another example relating to a method for establishing an emergency decision-making model for unmanned vehicles, which involves training the decision model using the gender and age of pedestrians¹, the invention’s model differentiates between objects to be protected and objects to be hit based on the pedestrian's gender and age. The CNIPA holds that AI and big data should avoid algorithmic discrimination, must conform to ethical and moral standards, and must not set algorithmic rules that are discriminatory or biased. This invention may involve social prejudice in its model that is against the Article 5 of the Chinese Patent Law and should not be granted.

 

III. Eligible Subject Matter

The subject matter eligibility discussed here refers to whether the technical subject matter sought for patenting complies with the provisions of Article 2 of the Chinese Patent Law regarding patent subject matter. If an invention does not involve a “technical problem,” a “technical means,” or a “technical effect,” it does not fall under eligible subject matter. For example, data obtained using AI technology may be highly significant for R&D activities, but it cannot be directly protected by a patent in China. Patent protection may be obtained for a technical solution that utilizes such data to solve a specific technical problem and achieve a corresponding effect.

 

In an example concerning a method for measuring sugar content at the initial stage of apple fruiting and using this data to predict the sugar content at maturity², the inventor measured the sugar content data at the initial stage of apple fruiting, input this data into an AI model for analysis, obtained the predicted sugar content at apple maturity, and adjusted the apple tree cultivation method based on this result. In this example, the "sugar content data at the initial stage of apple fruiting" itself does not solve any technical problem and thus does not belong to an eligible subject matter under the Chinese patent law. However, if there is a technical solution of transforming this data into predicted apple maturity sugar content data via AI technology and subsequently improving the apple tree cultivation method, the technical solution solves the technical problem of being unable to adjust the cultivation method before the apples ripened this season, achieving the effect of producing mature apples with high sugar content. Therefore, this technical solution is an eligible subject matter and complies with the provisions of Article 2 of the Chinese Patent Law.

 

IV. Sufficient Disclosure

Article 26, paragraph 3 of the Chinese Patent Law stipulates: “The description shall provide a clear and complete description of the invention or utility model so that a person skilled in the relevant field of technology can carry it out.” This requirement asks the applicant to disclose the technology of the invention in the patent description in a sufficient level, so that others can also implement the invention. However, AI technology often has a "black box" characteristic, making it difficult for the public to understand its internal structure and operating mechanism. CNIPA has recognized this point, and thus the Decision specifically addresses the standard of sufficient disclosure that the description of a patent involving AI technology should meet.

 

The Decision requires that the description should disclose the modular composition, hierarchical design, or connection relationship that defines the core framework of the model, as well as the specific steps and parameters of the training that embodies the optimization method. For inventions involving the specific application of AI, the description should clearly state how the model or algorithm is combined with the specific field or scenario, and how the input and output data of the algorithm or model are set.

 

In an example where the technology relates to how to improve the accuracy of malignant tumor prediction¹, the invention utilizes a trained malignant tumor enhanced screening model, taking routine blood tests, blood biochemical testing indicators, and facial image features collectively as the input for the screening model, in an attempt to obtain the predicted value of malignant tumor incidence. However, the description neither specifies which specific indicators are key to the accuracy of tumor prediction, nor states whether all indicators were referenced and given different weights for prediction. At the same time, based on current scientific research, except for a few types of tumors like facial skin cancer, it is uncertain whether there is an association between facial features and the incidence of malignant tumors. The description also does not state or prove the causal relationship between the “basis factors for judgment” and the “result of the judgment.” The description does not provide any verification data to prove that the accuracy rate of the prediction. Therefore, the technical solution has not been sufficiently disclosed in the description.

 

V. Novelty and Inventiveness

When examining the novelty and inventiveness of a patent application, the examiner will comprehensively consider all technical features described in the claims, including both non-AI technical features and AI technical features. If the difference from the prior art lies only in the use of AI technology, it is necessary to consider whether such use is common general knowledge. For example, an invention differs from the prior art in that previous R&D data is input into a neural network to predict results. However, in the field of machine learning, the method of using a trained neural network to predict future results by inputting historical data is known. Therefore, such a difference will not be sufficient to bring novelty and inventiveness to the invention.

 

In an example relating to a method for classifying scrap steel grades using AI technology¹, the differences between this method and the prior art include: the training data and the extracted features were different, and the number of lines and the hierarchical settings of the convolutional layers and pooling layers were also different. These differences were neither disclosed by the prior art nor belonged to common general knowledge. Therefore, the method possessed the novelty and inventiveness required by the Patent Law.

 

Conclusion

It is foreseeable that the number of AI patent filing in China will continue to grow in the coming years, and new issues will constantly emerge. CNIPA’s pace in updating its examination standards is also accelerating. Ling Chuang team will continuously monitor the latest developments in this hotspot and timely share our latest insights with our clients and industry peers. For any questions regarding this article and Ling Chuang Law Firm, please contact:

Mr. Yue (Pat) Sun, Co-Director, Patent Attorney, Attorney-at-Law, suny@lingchuanglaw.com;

Mr. Bin Shen, Manager, Patent Attorney, Attorney-at-Law, binshen@lingchuanglaw.com

 

References:  

1: Decision on Amending the Guidelines for Patent Examination (CNIPA Order No. 84), November 13, 2025.

2: Comparative Study Report on AI Patent Examination Cases in China and Japan, 2023, CNIPA and JPO.

 

About Ling Chuang: Beijing Ling Chuang Law Firm was established in January 2017 and is a professional legal service institution focusing on intellectual property services while also covering various types of dispute resolution. The core members of the firm’s team hold both lawyer and patent attorney qualifications, and have years experiences in patent and trademark application, litigation, and other facets of IP. For more information about Ling Chuang, please visit our website: http://www.lingchuanglaw.com/en/