The China National Intellectual Property Administration (CNIPA) announced the decision to amend the Patent Examination Guidelines on November 13, 2025. The revised guidelines will officially take effect on January 1, 2026. This comprehensive revision covers a wide

range of aspects, including procedural matters such as the provision and verification of inventor identity information, handling of same-day applications, and priority claims, as well as the refinement of examination rules for invention patent applications involving artificial intelligence (AI) and bitstreams, among others. This article provides a brief interpretation and commentary on the contents of this revision.

I. Requirements for Inventor Identity Information and Responsibility of Patent Agencies

This revision raises the requirements for inventors’ identity information and adds the verification responsibility of patent agencies.

1.1 Filling Out and Examination of Inventor Identification

Section 4.1.2, Chapter 1, Part I of the revised Guidelines clarifies that the inventor shall be a natural person. The identity information of all inventors shall be filled in the request form, and the authenticity of such information shall be ensured. Meanwhile, the revised Guidelines emphasize prohibition of listing fictitious inventors and establish a basis for examination: the examiner may proactively examine this information when there is evidence indicating that the listed inventors fail to comply with the provisions.

The above-mentioned revisions serve a dual purpose: on the one hand, they offer further clarification that AI cannot be listed as an inventor, and eliminate the possibility of listing AI as an inventor in practice. On the other hand, they establish regulations against

abnormal patent application behaviors characterized by false inventor information.

In current practice, the request form usually requires the names of all inventors. For the first inventor, nationality information shall be provided, and if the first inventor is Chinese, personal identification information such as the ID card number is also needed. 

This revision raises the requirements for inventor identity information. While the specific identification details required after the revision takes effect still await further clarification from the CNIPA, it is estimated, based on existing practice, that they should include at least the nationality information of all inventors. Applicants should prepare and verify the identity information of every inventor in advance to avoid subsequent legal risks arising from inaccurate information.

1.2 Information Verification Responsibility of Patent Agencies

Section 4.1.6, Chapter 1, Part I of the Guidelines adds a new provision, requiring that 

“a patent agency shall verify the applicant’ s identity information and contact information filled in the request form.” 

It is noteworthy that, compared with the expression in the draft for public consultation—which required the agency "to ensure the authenticity and validity of the inventor's/applicant's identity information and contact details"—the final version limits the agency's duty to merely "verification" of the applicant's information. This change relatively lowers the responsibility imposed on patent agencies. 

II. Notification Regarding Failure to Claim Priority in a Divisional Application

This revision supplements the following provisions in both Sections 6.2.1.2 and 6.2.2.2, Chapter 1, Part I of the Guidelines: 

“If the parent application of a divisional application claims priority, but the applicant fails to declare the priority claim in the request document when filing the divisional application, the divisional application shall be deemed not to have claimed the priority, and the examiner shall issue a Notification of Deemed Non-Claim of Priority.”

Under current practice, if an applicant fails to declare the priority claim of the original application due to negligence when filing a divisional application, the Patent Office usually does not take the initiative to issue a notice but directly deems that the applicant has waived such right. Unaware of this situation, the applicant may be unable to take timely remedial measures to restore the priority.

However, according to the new provision, the examiner must issue a “Notification of Deemed Non-Claim of Priority” in the above circumstances to alert the applicant of the omission. 

This allows the applicant to request the restoration of the priority right pursuant to Section 6.2.6.1, Chapter 1, Part I of the Guidelines. This section stipulates that if the parent application of a divisional application had claimed priority, but the divisional application is deemed not to have claimed priority due to certain reasons, the applicant may request for the restoration of that right.

This change provides an important procedural guarantee for applicants to claim priority. The examiner’ s active notification enables the applicant to promptly initiate the right restoration procedure after learning of the situation, thereby effectively preventing the loss of the priority right they should enjoy due to non-substantive oversight.

III. Adjustments to the Handling of Same-Day Applications

The revision modifies the rules for handling cases where "an applicant files on the same day (the date of filing) applications for both a utility model patent and an invention patent relating to the identical invention-creation" (hereinafter, "same-day applications"). The adjustment is made in Section 6.2.2, Chapter 3, Part II of the Guidelines.

For the same-day applications, the new Guidelines stipulate that the applicant shall separately declare at the time of filing that another patent application is filed for the identical invention-creation. If the applicant fails to fulfill this declaration obligation, the application shall be handled in accordance with the principle of Article 9, Clause 1 of the Patent Law, which states that “only one patent right shall be granted for the identical invention-creation.” 

In such cases, the applicant cannot obtain the invention patent right by declaring the abandonment of the utility model patent right. The applicant must amend the claims of the invention patent application to make the technical solution of the claims different from the granted utility model patent. This action is necessary to secure the invention patent grant, and to ensure the invention patent is not subsequently declared invalid later for violating Article 9, Clause 1 of the Patent Law.

On the other hand, if the applicant has made the declaration in accordance with the relevant provisions, a different procedure applies:

When the invention patent application is found to have no grounds for rejection after examination, the applicant must declare the abandonment of the granted utility model patent right within the specified time limit to obtain the invention patent right.

If the applicant refuses to abandon or fails to respond within the time limit, the invention patent application shall be rejected or deemed withdrawn. This means that once the declaration is made, the applicant can no longer retain both patent rights by amending the

claims of the invention application.

The original intention of the same-day application system is to allow applicants to enjoy the dual advantages of the fast authorization of utility model patents and the long protection period of invention patents. Specifically, the system allows applicants to first obtain rapid protection through the utility model patent and then "replace" it with the invention patent, which has a longer protection period, by abandoning the utility model patent right after the invention patent passes substantive examination.

Overall, the new provisions are more in line with the original intention of the same-day application system and consistent with the provisions of Rule 47 of the newly revised Implementing Regulations of the Patent Law in 2023.

This revision puts forward clearer requirements for applicants in choosing application strategies. Applicants should make prudent decisions when formulating their layouts: 

• If the aim is to ultimately obtain the invention patent right through “replacement,” they must fulfill the declaration obligation when filing the application. 
• If they wish to retain both the utility model patent right and the invention patent right, they shall not make the same-day application declaration and must ensure that there are substantive differences in the scope of claims between the two applications

IV. Refinement of the Principles for Examining Inventive Step

The revision amends Section 6.4 of Chapter 4, Part II of the Guidelines, and the core amendment adds a principle: 
“The feature not contributing to the solution of technical problems, even if included in the claims, usually does not affect the inventive step of the technical solution.” 

To illustrate this principle, the Guidelines provide an examination example. The example describes an invention related to a camera, where the core technology lies in improving the internal mechanical and circuit structures to solve the technical problem of “how to

achieve more flexible control of the shutter.” In response to the inventive step deficiency pointed out in the examination opinion, the applicant added conventional features such as the shape of the camera housing, the size of the display screen, and the position of the

battery compartment to the claims. However, since the specification neither clarifies any connection between these newly added features and the solution to the technical problem, nor could the applicant provide evidence or sufficient reason to prove that these

features bring any further technical effect to the technical solution, these added features are not considered in the inventive step judgment.

This revision clarifies and strengthens a key examination concept: simply stacking features unrelated to solving the core technical problem in the claims cannot endow the overall solution with inventive step. This is consistent with the current examination practice of focusing on the substantive contribution of the invention. 

The positive implications of this change are:

• Enhanced Examination Efficiency: It allows the examiner’ s judgment to better focus on the substantive contribution of the invention, reducing debate over non-relevant features.
• Clarified Judgment Standards: It provides clearer criteria for determining “whether a technical feature affects the inventive step,” which helps reduce subsequent disputes.

For applicants and patent attorneys, this revision puts forward higher requirements. When drafting application documents, it is recommended to emphasize the technical effects brought by technical features—especially those that may constitute an inventive point but

are not included in the initial claims. Clearly and fully elaborating their technical contributions in the specification will lay a solid foundation and provide stronger support for the subsequent examination procedures, especially when arguing the inventive step of the invention.

V. Improvement and Refinement of Examination Standards for Inventions Related to AI and Big Data

This revision of the Guidelines improves and refines the examination standards for invention patent applications containing algorithms or business rules and methods features, such as those related to AI and big data. The changes cover multiple dimensions, including

ethical examination, inventive step judgment, and requirements for sufficient disclosure in the specification. The relevant modifications are mainly concentrated in Section 6, Chapter 9, Part II of the Patent Examination Guidelines.

5.1 Ethical Examination for AI-Related Patent Applications

Article 5, Clause 1 of the Patent Law stipulates that no patent right shall be granted for any invention-creation that violates laws, social morality, or that is detrimental to the public interest. Due to its deep dependence on data resources, the development and application

of AI technology are often accompanied by potential risks of algorithm ethics and data compliance. However, prior to this revision, the Guidelines did not detail how Article 5 should specifically apply to AI-related applications.To address this demand, the revision adds Section 6.1.1 to Chapter 9, Part II, specifically regulating the ethical examination of AI-related applications. Additionally, two typical examples are supplemented in “6.2 Examination Examples” to enhance the operability of examination practice.

The new Section 6.1.1 stipulates: 

“For invention patent applications including algorithm features or business rules and method features, if the data collection, label management, rule setting, or recommendation decision-making contains content that violates laws, social morality, or harms public

interest, the patent right cannot be granted according to Article 5, Clause 1 of the Patent Law.” 

The two new examples illustrate situations where patent rights cannot be granted, based on violating laws and violating social morality, respectively.

Example 1: The claim involves “A sales assistance system for mattresses in shopping malls based on big data.” To achieve precise marketing, the system collects customers’ facial features to extract identity information. This operation violates the provisions on personal information processing in the Personal Information Protection Law of the PRC, so the claim does not meet the requirements of Article 5, Clause 1 of the Patent Law.

Example 2: The claim involves “A method for establishing an emergency decision-making model for an unmanned vehicle.” The method uses the gender and age of pedestrians as obstacle data to determine the“protected object” and the “collided object” in unavoidable collisions. This solution violates social ethics and morality by discriminating life value based on natural attributes like gender and age, and therefore cannot be granted a patent right.

In practice, when drafting AI-related patent applications, applicants must handle all descriptions involving personal information cautiously to ensure the legality of the entire process. China has established a legal system for personal information protection, including the Civil Code, the Personal Information Protection Law, the Data Security Law, etc. Violation of these norms will trigger the prohibitive provisions of Article 5 of the Patent Law. Furthermore, applications involving the specific field application of algorithms must also comprehensively scrutinize the risk of violating the law, social morality, or public interest, ensuring the legality and ethics of the claimed technical innovation.

Notably, Section 6.1 in this chapter also clarifies that the scope of patent examination is not limited to the claims but should also include the content of the specification when necessary. This requires applicants to not only ensure the legality of the claims but also to avoid documenting any content that violates the law or public order or morality in the specification.

5.2 Further Clarification on Inventiveness Determination Standards

To refine the standards for judging the inventive step of inventions in fields like AI, two new examples are added to “6.2 Examination Examples” in Chapter 9, Part II of the Guidelines.

New Example 18: Lack of Substantial Improvement in Algorithm Application

This example is from the Guidance for Patent Applications Related to Artificial Intelligence (Trial) published at the end of 2024 (hereinafter, the Guidance). The claim relates to a method for identifying the number of ships, while the cited reference discloses a method for

identifying the number of fruits on trees. The only difference is the identified object. Since no substantive adjustments were made to the deep learning model's training method, structure, or parameter configuration to overcome specific problems in the new scenario,the claim is deemed to lack inventiveness.

This example reflects the principle for judging inventive step stated in the Guidance: “If an algorithm or model is merely applied to a different scenario without overcoming technical difficulties by adjusting elements like the training method, parameters, orconfiguration, and does not produce an unexpected technical effect, the solution lacks inventiveness.”

New Example 19: Model construction with specific improvements

This case is one of the Top 10 Reexamination and Invalidation Cases of 2023. The claim involves “A method for establishing a neural network model for grading scrap steel.” The difference from cited Reference 1 - a general training method for scrap steel type identification models - lies in the adoption of different training data andfeature extraction methods, as well as the specific adjustment of the number of paths and layer settings of convolutional and pooling layers. The combination of these distinctive algorithmic features and technical features result in improved accuracy in scrap steel grade classification. Therefore, the claim is deemed to possess inventiveness.

This case demonstrates that inventiveness can be recognized when the solution is not a simple transplant of an existing model or algorithm, but rather a targeted improvement—capable of bringing beneficial technical effects—to core elements like data being processed and model structure, based on the unique requirements of a specific application scenario.

5.3 Further Refinement of the Sufficient Disclosure Requirement

To address the particularities of patent applications in the field of AI, this revision makes more targeted provisions for the sufficiency of disclosure: 

“If the application involves the construction or training of AI models, the specification generally needs to clearly record the necessary modules, layers, or connection relationships of the model, as well as the specific steps and parameters required for training; If it involves applying an AI model or algorithm in a specific field or scenario, the specification generally needs to clearly record 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 to indicate their inherent correlation, so that a person skilled in the art can realize the solution of the invention based on the content recorded in the specification.”

In addition, the revision uses two examples (positive and negative) in the newly added Section 6.3.3 to further illustrate the examination standards.

Positive Example (Example 20) involves “A method for generating facial features.” The claim defines that a spatial transformation network is set in the first convolutional neural network, but the specification does not clearly specify the specific position of the spatial transformation network in the first convolutional neural network. The analysis holds that since both network structures are well-known in the field, and a person skilled in the art knows that the spatial transformation network can be inserted as a whole at any position of the convolutional neural network, the specification is deemed to have sufficiently disclosed the invention.

Negative Example (Example 21) involves “A method for predicting cancer based on biological information.” The solution uses a trained malignant tumor enhanced screening model, taking routine blood/biochemistry indicators and facial image features as input to predict the incidence of malignant tumor. However, the correlation between facial features and the incidence of malignant tumors is uncertain. The specification does not record or prove the causal relationship between the “judgment basis factors” (e.g., facial features) and the“judgment result,” and fails to provide evidence that the model can improve the prediction accuracy compared with traditional cancer biomarkers. Therefore, the specification is deemed insufficiently disclosed.

As pointed out in the Guidance, AI algorithms or models inherently have the characteristics of a “black box,” requiring sufficient information to achieve full disclosure. While rejections for insufficient disclosure have been relatively few thus far, the direction set by this Guidelines revision and the Guidance suggests that sufficiency of disclosure will become a future examination priority for AI-related patent applications.

It should be particularly noted that if an application document is deemed to have the defect of insufficient disclosure, it is often difficult to overcome through amendment during the examination stage. This requires applicants to attach great importance to this issue during the drafting stage, clearly and completely describe the core content closely related to the technical contribution, thereby ensuring it meets the standard of enabling a person skilled in the art to implement the invention.

VI. Provisions on the Examination of Patent Applications Involving Bitstreams

To adapt to the rapid development of streaming media technology and the continuous evolution of application scenarios, and to address the practical needs of innovation entities for strengthened patent protection across the streaming media industry chain, the revision adds Section 7: Provisions on the Examination of Patent Applications Involving Bitstreams" in Chapter 9, Part II. This section clarifies specific requirements for determining patentable subject matter and drafting the specification and claims for this type of application.

6.1 Subject Matter Eligibility
Regarding patentable subject matter, the revision explicitly states that if the subject matter of a claim involves only a mere bitstream, or if all its limitations (besides the subject title) point only to the bitstream itself, it falls under the category of “rules and methods for mental activities” and does not constitute patentable subject matter.

For example, the following two types of claims are not considered patentable:

1. A bitstream, comprising syntax element A, syntax element B, ...

2. A method for generating a bitstream, wherein the bitstream comprises syntax element A, syntax element B, ...

6.2 Specification Drafting Requirements

Regarding the drafting of the specification, Section 7.2.1 requires that the specification must provide a clear and complete description of the specific video coding method used to generate the bitstream.

6.3 Claim Drafting Requirements
Regarding the drafting of the claims, Section 7.2.2 stipulates that for applications involving bitstreams generated by a specific video coding method, the claims can be constructed as various types, including, e.g., video encoding /decoding methods and apparatus,

storage and transmission methods , and computer-readable storage media. Generally, the core claim is recommended to be the specific video encoding method claim that generates the bitstream, from which other claims can be derived. The revision provides the following examples:

1. A video encoding method comprising: a frame partitioning step, ..., an entropy encoding step, ...

2. A video encoding apparatus comprising: a frame partitioning unit, ..., an entropy encoding unit, ...

3. A video decoding method comprising: an entropy decoding step, ..., a frame output step, ...

4. A video decoding apparatus comprising: an entropy decoding unit, ..., a frame output unit, ...

5. A method for storing a bitstream comprising: generating the bitstream by executing the video encoding method of claim 1; and storing the bitstream.

6. A method for transmitting a bitstream comprising: generating the bitstream by executing the video encoding method of claim 1; and transmitting the bitstream.

7. A computer-readable storage medium, having stored thereon a computer program/computer instructions and a bitstream, the computer program/instructions, when executed by a processor, performing the video encoding method of claim 1 to generate the bitstream.

According to the above provisions, a mere data stream does not constitute patentable subject matter, which is consistent with the current examination standards regarding subject matter eligibility. The provisions offer a clear path for patent strategy in the field of digital video coding and decoding: to achieve indirect protection for a data stream, applicants should focus on the specific video coding method as the core and build different types of claims based on it, to forming a systematic patent protection network.

VII. Modifications to Patent Invalidation Procedures

The revision mainly covers three aspects concerning patent invalidation procedures.

7.1 Strengthening the Verification of Petitioner Authenticity

The revision adds a new condition for non-acceptance in Section 3.2 "Qualification of the Petitioner for Invalidation," Chapter 3, Part IV: "where the request for invalidation is not a true expression of the petitioner's 
intent." 

Concurrently, Section 4.1.6, Chapter 1, Part I is supplemented with the following provision: "If a patent agency or patent attorney files a patent application or requests patent invalidation in its/his/her own name, it shall be dealt with in accordance with the Patent Agency Regulations."

These revisions aim to address improper acts emerging in practice, such as filing invalidation requests under unauthorized names, or patent attorneys filing through the names of friends or relatives. This change institutionally strengthens the examination requirements

for the authenticity of the petitioner. However, based on common understanding, an invalidation filed in the name of a "straw man" should not fall under the scope of non-acceptance if the "straw man" is aware and expresses genuine intent.

7.2. Clarifying Requirements for Amending Claims

Section 4.6.4, Chapter 3, Part IV of the revision further clarifies that when the patentee amends the claims in invalidation proceedings, the patentee shall submit both the full replacement pages and the comparison pages concurrently.

Furthermore, if multiple compliant amended texts are submitted within the same invalidation proceeding, the last submitted text shall be used as the basis for examination.

This regulation clarifies the formal requirements and procedural rules for amended texts, which helps to enhance the efficiency and operability of invalidation proceedings.

7.3. Clarifying the Principle of Prohibition on Repeated Request

Section 3.3, Chapter 3, Part IV of the revision further clarifies the applicability of the "Prohibition on Repeated Request" principle: A request for invalidation filed again against a patent right for which an invalidation decision has already been rendered shall not be accepted if it uses the "same or substantially the same" grounds and evidence.

For example, making only simple formal adjustments or changes to the invalidation grounds or evidence, while the underlying legal facts remain substantially the same as the previous request, still falls under the scope of the "Prohibition on Repeated Request" principle. This revision aligns with current examination practice, enhancing the stability and predictability of legal application.

VIII. Regarding Patent Term Adjustment (PTA)

The revision further expands the scope of "reasonable delays during the granting process" (i.e., circumstances excluded from the PTA calculation). It is specifically clarified that during reexamination, even without amendments to the application documents, the time spent is considered a reasonable delay if the rejection decision is revoked based on new arguments or new evidence submitted by the petitioner.

Given that petitioners are less likely to file a request for reexamination based on their original arguments and evidence, this change effectively excludes the entire reexamination period for most cases from the PTA calculation.

IX. Other Amendments

In addition to the aforementioned changes, other important adjustments in the revision also include:

- The definition of plant varieties is updated.

- For the purpose of calculating the excess pages fee, sequence listings submitted in a computer-readable format are no longer included in the total number of pages.

- Regarding refunds, all circumstances where the Patent Office currently issue refunds on its own initiative have been adjusted to require a request from the applicants.

The revision of the Patent Examination Guidelines actively addresses the demands for patent protection brought about by emerging technologies such as AI and big data. At the same time, by refining examination standards and procedures, it aims to improve the quality and efficiency of patent examination. For applicants, understanding and adapting to these new regulations will help them better protect their innovations and enhance their competitiveness in intellectual property.