Richard Yong WANG  Partner, Attorney-at-law and Patent Attorney

 

The China National Intellectual Property Administration (CNIPA) issued Order No. 84 on November 10, 2025, revising the Patent Examination Guidelines, which comes into effect on January 1, 2026. This revision includes significant additions to Chapter 9 of Part Two of the Patent Examination Guidelines（PEG） regarding the examination of invention patent applications concerning computer programs in the fields of artificial intelligence, big data, and bitstreams, particularly in the section on subject matters examination. This article provides an interpretation of the newly added content for the reference of inventors and applicants.

 

Specifically, a new Section 6.1.1 has been added to Chapter 9 of Part Two of the PEG. The added content is excerpted as follows (all underlined content below is quoted from the amended PEG):

 

6.1.1 Examination under Article 5.1 of the Patent Law

 

For invention patent applications containing algorithmic features or business rule and method features, if elements such as data collection, label management, rule setting, or recommendation decisions involve content that violates laws, social morality, or is detrimental to the public interest, then according to the provisions of Article 5.1 of the Patent Law, a patent right cannot be granted.

 

Correspondingly, in Section 6.2 (Examination Examples) of this chapter, two examination examples for invention patent applications under Article 5.1 of the Patent Law have been added. The added examples and their corresponding analyses are as follows:

(1) Invention patent applications containing algorithmic features or business rules and method features, if they violate laws, social morality, or are detrimental to the public interest, cannot be granted a patent right.

 

【Example 1】

An In-store Mattress Sales Assistance System Based on Big Data

Summary of Invention:

The solution described in the invention patent application is a big data-based in-store mattress sales assistance system. It utilizes camera modules and facial recognition modules to collect customers' facial feature information and obtain their identity recognition data. The collected information is then subjected to data analysis to assess customers' genuine preferences for mattresses, thereby assisting merchants in precision marketing.

Claims:

A big data-based in-store mattress sales assistance system, comprising a mattress display device and a management center, characterized in that:

The mattress display device includes a control module and an information collection module, used to display and assist in the sale of mattress products while collecting customer data; the control module is used for data interaction with the management center; the information collection module includes a camera module and a facial recognition module, used to collect customers' facial feature information, adjust facial posture using a keypoint detection algorithm to obtain a normalized facial image, locate the facial region to be recognized through a facial detection algorithm on the normalized facial image, and extract facial features within the facial region using Principal Component Analysis (PCA), thereby obtaining customer identity recognition information;

The management center comprises a management server and an analysis assistance system; the management server is used to manage multiple mattress display devices; the analysis assistance system analyzes the data collected by the mattress display devices based on the customer's identity recognition information to derive the customer's genuine preferences and feeds back the analysis results to the management center.

 

Analysis and Conclusion

Relevant provisions of the Personal Information Protection Law of the People's Republic of China stipulate that the installation of image capture and personal identity recognition equipment in public places must be necessary for maintaining public safety, comply with relevant state regulations, and be accompanied by prominent warning signs. The collected personal images and identity information may only be used for the purpose of maintaining public safety and shall not be used for other purposes unless separate consent is obtained from the individual.

As evident from the solution claimed in this invention, the use of image capture and facial recognition means in business premises such as shopping malls for precision marketing of mattresses does not constitute a necessity for maintaining public safety. Furthermore, to obtain and analyze customers' genuine preferences for mattresses, the collection of customers' facial information and acquisition of their identity recognition information are clearly conducted without the customers' awareness, and the application fails to indicate that the data acquisition or information collection is lawful or compliant also. Therefore, this invention contravenes the law and, according to Article 5.1of the Patent Law, cannot be granted a patent right.

 

【Example 2】

A Method for Establishing an Emergency Decision-Making Model for Unmanned Vehicles

Summary of Invention:

The solution described in the invention patent application is a method for establishing an emergency decision-making model for unmanned vehicles. It uses the gender and age of pedestrians as obstacle data and employs a trained decision-making model to determine the protected target and the collision target in situations where obstacles cannot be avoided.

Claims:

A method for establishing an emergency decision-making model for unmanned vehicles, characterized by comprising:

acquiring historical environmental data and historical obstacle data of the unmanned vehicle, wherein the historical environmental data includes the vehicle's travel speed, distance to obstacles in its own lane, distance to obstacles in adjacent lanes, motion speed and direction of obstacles in its own lane, and motion speed and direction of obstacles in adjacent lanes; and the historical obstacle data includes the gender and age of pedestrians;     

Performing feature extraction on the historical environmental data and historical obstacle data to serve as input data for the decision-making model; using the historical travel trajectories of the vehicle when obstacles cannot be avoided as the output data for the decision-making model; and training the decision-making model based on the historical data, wherein the decision-making model is a deep learning model;
   Acquiring real-time environmental data and real-time obstacle data; when the unmanned vehicle encounters a situation where obstacles cannot be avoided, utilizing the trained decision-making model to determine the travel trajectory of the unmanned vehicle.

 

Analysis and Conclusion

This invention involves a method for establishing an emergency decision-making model for unmanned vehicles. Human life possesses equal value and dignity regardless of age or gender. If an emergency decision-making model for unmanned vehicles, in unavoidable accident scenarios, selects individuals to protect or collide with based on the gender and age of pedestrians, it contradicts the public ethical belief in equality of all lives. Furthermore, such a decision-making approach would reinforce existing societal biases related to gender and age, raise public concerns about the safety of transportation, and undermine trust in technology and social order. Therefore, this invention contains content that goes against social morality and, according to Article 5.1 of the Patent Law, cannot be granted a patent right.

 

 

Author’s Comments：

 

Regarding the examination criteria under Article 5.1 of the Patent Law, general provisions were already outlined in Section 3 of Part Two, Chapter 1 of the PEG prior to this revision, including: 3.1.1 Inventions that violate laws, 3.1.2 Inventions that violate social morality, and 3.1.3 Inventions that are detrimental to the public interest.

 

The reason for dedicating a separate section in Chapter 9 Part Two to this revision, specifying how to examine the subject matter of inventions involving artificial intelligence and big data in accordance with Article 5.1 of the Patent Law, is due to the rapid development of technologies such as AI and big data. This development has led to the emergence of numerous new algorithms and models, giving rise to diverse applications across various industries. Meanwhile, the application of these technologies relies heavily on large amounts of data and information resources, and their in-depth development inevitably brings forth ethical issues such as algorithmic ethics, data security, and data compliance. For instance, questions may arise when solutions involving algorithms or models are applied to specific fields regarding potential violations of relevant laws, social morality, or harm to the public interest. Regarding AI's acquisition and use of data, issues must be examined concerning whether each stage—from data sources and application scenarios to security management and usage norms—complies with relevant laws. Furthermore, beyond the data content itself, specific methods of data collection, storage, and processing must be assessed for compliance with legal requirements and whether they violate social morality or harm the public interest. Therefore, it is necessary to refine, revise, and improve the examination provisions in this area alongside technological advancements.

 

Regarding【Example 1】

【Example 1】involves a big data-based mattress sales assistance system for use within shopping malls. It encompasses processes such as image capture, facial recognition, and identity verification in public spaces. Since it involves the collection, recognition, and processing of personal information, it must comply with the stipulations of the Personal Information Protection Law of the People's Republic of China (hereinafter referred to as the "PIPL"). This invention conducts image capture and recognition within shopping malls, which qualify as public places. Therefore, the installation of image capture and personal identity recognition equipment, as well as the processing of the information collected, must adhere to the provisions of Article 26 of the PIPL. Although this invention does not directly involve the process of installing such equipment in public places, if it utilizes illegally installed devices in public settings to collect and obtain personal information, the very act of information collection is unlawful. Consequently, the subsequent use and processing of such collected information are also illegal. Furthermore, Article 26 of the PIPL also mandates that personal images and identification information collected through lawfully installed image capture and personal identity recognition equipment in public places may only be used for the purpose of maintaining public safety and shall not be used for any other purposes.

 

The invention in 【Example 1】 also involves facial recognition and identity verification, a process that pertains to sensitive personal information. According to Article 28 of the PIPL, sensitive personal information includes, among others, biometric data, religious beliefs, specific identity details, medical health information, financial accounts, location trajectories, and the personal information of minors under the age of fourteen. When processing sensitive personal information, Article 29 of the PIPL stipulates that separate consent from the individual must be obtained; under specific circumstances, written consent is even required. Therefore, based on the aforementioned provisions, the invention in 【Example 1】, which collects facial information and obtains identity recognition data of customers in public places—without demonstrating that the data acquisition or information collection is lawful, compliant, or necessary for maintaining public safety—clearly contravenes the law and cannot be granted a patent right.

 

China's "Interim Measures for the Management of Generative Artificial Intelligence Services" also stipulate that when providers engage in training data processing activities such as pre-training and optimization training involving personal information, they must obtain individual consent or comply with other circumstances specified by laws and administrative regulations. Providers are required to fulfill their protection obligations in accordance with the law regarding users' input information and usage records. They must not collect unnecessary personal information, illegally retain input information and usage records that can identify users, or unlawfully provide users' input information and usage records to others.

 

However, it should be noted that, according to Article 10 of the Implementing Regulations of the Patent Law, the term "invention-creation that violates the law" referred to in Article 5 of the Patent Law does not include those invention-creations only the implementation of which is prohibited by law. That is to say, if only the production, sale, or use of a product resulting from an invention-creation is restricted or prohibited by law, neither the product itself nor its manufacturing method constitutes an invention-creation that violates the law. The PEG provides an example regarding weapons. Although the production, sale, and use of weapons are restricted or prohibited by law, weapons themselves and their manufacturing methods are still considered the eligible subject matters for patent protection. Therefore, in the case of invention-creations such as [Example 1], even if the technical solution involves image collection, biometric feature extraction, identity recognition, etc., as long as the solution containing such technical content itself does not violate the law—and it is only the implementation of the technical solution that is restricted or prohibited by law due to the need to collect personal information or conduct identity recognition (such as the provisions in the PIPL regarding the prohibition of collecting personal information without authorization, or obtaining facial features and identity recognition without permission)—the examiner should not reject the corresponding invention-creations under Article 5.1.

 

In order to avoid violating laws and regulations such as the PIPL, patent applications involving the processing of personal information should clearly specify in the application document how to ensure the legality of data acquisition and usage. For example, if the invention involves large-scale image collection and processing in public places, the purpose of the invention such as maintaining public safety should be clearly stated. For the collection and use of information or images, measures must be taken to ensure that personal privacy is not infringed upon. This can include obtaining individual explicit consent or deemed consent, or removing sensitive personal information through de-identification (i.e., processing personal information so that it cannot be linked to a specific individual without the use of additional information) or anonymization (i.e., processing personal information so that it cannot be linked to a specific individual and cannot be restored). Finally, for patent applications involving data (especially personal information) or data processing, the applicant or agent may include a statement in the patent application specification regarding information collection and processing, for example: "The user information and data involved in this application are all authorized by users or fully authorized by relevant parties.", etc. This can help alleviate any concerns the examiner may have regarding this issue.

 

Regarding [Example 2]

[Example 2] involves emergency decision-making for driverless vehicles. This example is essentially similar to the "Trolley Problem." The Trolley Problem is a famous thought experiment in ethics. In essence, a runaway trolley is speeding down a track where five workers are laboring, unaware of the danger and soon to be killed. You are in a control room next to the track with a lever at hand: if you do nothing, the trolley will kill the five people; if you pull the lever, the trolley will switch to another track where only one worker is present. That one worker will be killed, but the five will be saved. Since its introduction in 1967, the Trolley Problem has become one of the most famous thought experiments in ethics. Over more than half a century, it has now evolved into one of the most pointed inquiries into modern technology, particularly the AI used in autonomous driving.

 

The ethical foundation widely recognized in modern society is the equality of all human lives. The life of every individual holds equal value and dignity, irrespective of social status, wealth, gender, or age. The decision-making model in the patent application of [Example 2], which incorporates gender and age as input data, essentially constructs a discriminatory mechanism for life-and-death choices. Such a mechanism violates the most fundamental societal morality that "all lives are equally precious." If such decisions were permitted, they would grant technological systems power beyond the ethical boundaries of humanity and explicitly acknowledge that the lives of certain groups (such as the elderly or specific genders) hold lesser value than others—an outcome that societal ethics absolutely cannot condone. This decision-making model is not only technical-related but also profoundly social-related. Moreover, encoding gender and age into life-and-death decisions would reinforce and solidify existing gender and age discrimination in society, which deviates from societal efforts to eliminate discrimination and promote equality through laws and policies, thereby harming the public interest and hindering the healthy development and social acceptance of technology.

 

Therefore, while the invention in [Example 2] provides a decision-making mechanism (a deep learning model) from a technical perspective, its input data (the gender and age of pedestrians) and decision-making objective (selecting who is protected and who is hit) directly lead to the following outcomes: 1) violating social morality by trampling on the ethical principle of the equality of all lives; 2) harming the public interest by undermining public trust in autonomous driving technology and exacerbating social discrimination and inequality. Consequently, in accordance with Article 5.1 of the Patent Law, this invention contains elements contrary to social morality and should not be granted a patent right.

 

Another example similar to [Example 2] is the application of AI in the medical field. The application of AI in the medical field is booming, from diagnostic image analysis to personalized treatment plans, and then to intelligent agent assisted decision-making. However, it also acts as a double-edged sword, amplifying humanity's most profound ethical dilemmas. If you bring the "trolley problem" of autonomous driving to the hospital, you may face a similar problem: who should an AI system prioritize for treatment when resources are limited? Is he a wealthy person or a city wanderer? Is it an elderly person or a child?

 

Although AI in medicine and autonomous driving technology will not halt progress due to such ethical challenges, inventions that contain content contrary to social morality or detrimental to the public interest receive a clear negative response within the framework of China's patent law and cannot be granted patent rights.

 

In summary, inventions involving artificial intelligence may give rise to the following legal and ethical issues: 1) Data Privacy and Protection: The training and operation of AI require large amounts of data, which may contain personal privacy information. During the collection, storage, use, and sharing of data, data protection laws and regulations must be complied with to ensure that the privacy of data subjects is not violated;. 2) Legal Sources of Data: The data used to train AI models must come from legal sources and must not infringe on the intellectual property rights or other legitimate rights and interests of others. As stipulated in the aforementioned Interim Measures for the Management of Generative Artificial Intelligence Services, providers of generative AI services are required to ensure that pre-training data is sourced legally and does not infringe on others' intellectual property rights. 3) Algorithmic Bias and Discrimination: Algorithmic bias often stems from biases in the training data. If the training data contains discrimination based on race, gender, age, or other factors, the AI model is likely to learn and amplify these biases and discrimination, leading to unfair outcomes. This may have adverse effects on specific groups and trigger legal disputes and social controversies. Therefore, when developing and drafting inventions related to AI products, close attention must be paid to the ethical and legal issues mentioned above.

 

Author:

 

 

Mr. Richard Yong Wang

 

Mr. Wang received his bachelor's degree in 1991 from the department of computer science of East China Normal University and his master's degree from the Institute of Computing Technology of the Chinese Academy of Sciences in 1994. In 2005, he received degree of master of laws from Renmin University of China. Mr. Wang joined Panawell in January 2007.

 

In the past years, Mr. Wang has handled thousands of patent applications for both domestic and foreign clients, and he has extensive experiences in application drafting, responding to office actions, patent reexamination and invalidation proceeding, patent administrative litigation, infringement litigation, software registration and integrated circuit layout design registration. As a very experienced patent attorney and attorney-at-law, Mr. Wang also participated in many patent litigation cases on behalf of a number of multinational companies as leading attorney. Mr. Wang's practices include computer hardware, computer software, communication technology, semiconductor devices and manufacturing process, automatic control, household electrical appliances, and etc.