On December 31, 2024, the China National Intellectual Property Administration (CNIPA) released the Guidelines for AI-related Invention Applications (Test Implementation), addressing several legal aspects, including inventor qualification/identification, for Artificial Intelligence related (AI-related) patent applications.

Below are some key highlights from the Guidelines.

Inventor qualification/Identification

The inventor(s) listed in a patent document must be a natural person or persons who have made creative contribution to the substantive features of the invention. Artificial Intelligence (AI) systems, which are not currently recognised as legal entities entitled to legal rights, are not eligible to be named as inventors.

Classification of AI-related invention patents

The Guidelines classify AI-related patents into the following four categories, among which categories 1 to 3 may be eligible for patent protection, while category 4 is not:

1. AI algorithms or models themselves, including their improvements or optimisations
2. Applications of AI algorithms or models for achieving specific functions or in specific fields
3. Patent applications assisted by AI, where AI functions as an auxiliary tool, similar to an information processor or a drawing tool
4. Applications generated autonomously by AI without human contribution.

Requirements on patent subject matters

Patent claims should include technical features associated with algorithmic features, utilising natural laws to address technical problems and produce technical effects that conform to the laws of nature.

For instance, in the claims,

1. the object processed by the AI algorithm or model should be data with definite technical meaning in the relevant technical field
2. the AI algorithm or model must have specific technical relationship with the internal structure of a computer system, thereby improving the internal performance of the computer system in accordance with the laws of nature, or
3. the inherent correlation among data generated through AI algorithms or models must conform to the law of nature.

Full disclosure in the specification

The specification should fully disclose the aspects of the invention that contribute to the state of the art.

For example, the specifications are suggested to include one or more of the following:

1. the training process and the specific steps of the algorithm and the training method, when the contribution lies in the training of an AI model
2. the necessary module structure, hierarchical structure, or connection relationship, as well as functions and effects of the model, when the contribution lies in the construction of an AI model
3. how to combine the model with the specific application scenario, how to set input/output data, etc., when the contribution lies in the application of AI in a specific field.

Inventiveness

In evaluating the inventiveness, the following algorithm features can be considered together with other technical features as a whole.

1. an existing AI algorithm or model with adjusted parameter(s), which solves the technical problems and achieves beneficial technical effect
2. an existing AI algorithm or model applied to a different scenario, which overcomes technical difficulty and/or brings unexpected technical effect, etc.
3. an AI ​​algorithm or model having specific technical connection with internal structure of a computer system, which improves internal performance of the computer system
4. AI ​​algorithm feature(s) for improving user experience.

Ethical considerations in AI-related patent applications

AI-related patent applications must comply with applicable social ethics and public interests. This includes ensuring that data sources, application scenarios, security management, usage specifications, and data collection, storage and processing methods conform to relevant laws.

How we can help

Located in Beijing and Hong Kong, our China IP team have deep expertise in Chinese patent law. Their strategic guidance is invaluable to our global and local clients, so please feel free to reach out to them with any questions relating to patents, patentability of AI inventions or the Guidelines for AI-related Invention Applications (Test Implementation).