The China National Intellectual Property Administration (CNIPA) has issued draft Guidelines for Patent Applications Related to Artificial Intelligence Inventions (Guidelines) to address the rapidly increasing volume of patent filings in artificial intelligence (AI). Released on 6 December 2024, the Guidelines aim to clarify patent examination policies under China’s existing patent law framework, providing better support to applicants navigating complex legal and technical issues. The public is invited to submit feedback on the draft by 13 December 2024. This initiative reflects the importance of AI as a driver of innovation and its alignment with China’s strategic focus on intellectual property protection in emerging technologies.
[Overview of the Guidelines]
The draft Guidelines span six chapters, covering key areas relevant to AI-related patent applications:
1. Common Types and Legal Issues
This chapter identifies four primary types of AI-related patent applications, each with distinct characteristics and challenges:
1) AI algorithms or models themselves, such as innovations in machine learning techniques, neural network architectures, or algorithm optimization methods. 2) Functional or domain-specific applications of AI models, such as AI-based diagnostic tools in healthcare or predictive maintenance systems in industrial settings. 3) Inventions created with AI assistance where AI serves as a tool aiding human inventors in tasks like material discovery or process optimization. 4) Inventions autonomously generated by AI, where human contribution is minimal or absent.
The chapter also explores pressing legal issues, including the recognition of inventorship, the adequacy of disclosure to meet legal standards, and the ethical implications of granting patents in such rapidly evolving contexts.
2. Recognition of Inventorship
The Guidelines explicitly state that inventors must be natural persons, affirming that AI systems cannot be designated as inventors under current laws. This position ensures adherence to existing legal frameworks and clarifies the roles of individuals in the inventive process. The chapter emphasizes that only those who contribute creatively to the core features of an invention can claim inventorship. For example, in cases where AI aids in generating ideas or solutions, the human’s role in directing, supervising, or refining the process is critical for establishing inventorship.
3. Subject Matter Standards
This section explains how AI-related patent claims must satisfy criteria for patentable subject matter. It clarifies that technical solutions leveraging natural laws, such as innovations in data processing methods or improvements in system efficiency, can qualify for protection. However, abstract rules, mathematical methods, or purely intellectual activities without a clear technical contribution are excluded. Detailed examples are provided to guide applicants in framing claims that highlight the technical aspects of AI innovations, such as incorporating algorithms into practical implementations like medical image analysis or autonomous vehicle systems.
4. Adequate Disclosure
The Guidelines emphasize the necessity of clear and complete disclosure in patent applications, particularly given the “black box” nature of many AI technologies. Applicants are required to provide sufficient details about the algorithms, training processes, data inputs, and outputs to enable others in the field to reproduce the invention. The chapter offers practical advice on how to document algorithmic steps, specify data requirements, and describe application scenarios comprehensively. These measures aim to ensure transparency and reproducibility, addressing common concerns about the opaque decision-making processes of AI systems.
5. Assessment of Inventive Step
This chapter provides detailed guidance on assessing the inventive step for AI-related patents. It explains how AI algorithms contribute to the technical solution of a problem, making them eligible for consideration in inventive step evaluations. Examples demonstrate scenarios where integrating AI algorithms with technical features enhances functionality or achieves unexpected results, such as improving energy efficiency in smart grids or optimizing resource allocation in computing systems. The section stresses the importance of showing how AI-specific features contribute to non-obvious technological advancements.
6. Ethical Considerations
The final chapter addresses the ethical implications of AI-related patents, offering principles to navigate concerns such as algorithm bias, data privacy, and public interest. It encourages applicants to ensure compliance with laws and ethical standards in areas like data usage, security, and the societal impact of AI applications. For example, the chapter highlights the importance of designing algorithms that are fair and unbiased while promoting responsible innovation that aligns with societal values. These considerations aim to foster the development of AI technologies that are both legally sound and ethically responsible.
[Comment]
The draft Guidelines issued by CNIPA provide a timely and comprehensive response to the needs of AI-related patent applications. The inclusion of detailed examples and scenario analysis is commendable, offering practical guidance to applicants in navigating examination policies. However, challenges remain in bridging the gap between abstract AI algorithms and concrete technical applications. While the emphasis on natural law-based technical solutions is clear, the subjective interpretation of technical contributions may lead to inconsistencies in examination. Further, while the inclusion of ethical considerations is important, balancing innovation incentives with restrictions will be crucial to avoid hindering technological progress.
[Overview of the Guidelines]
The draft Guidelines span six chapters, covering key areas relevant to AI-related patent applications:
1. Common Types and Legal Issues
This chapter identifies four primary types of AI-related patent applications, each with distinct characteristics and challenges:
1) AI algorithms or models themselves, such as innovations in machine learning techniques, neural network architectures, or algorithm optimization methods. 2) Functional or domain-specific applications of AI models, such as AI-based diagnostic tools in healthcare or predictive maintenance systems in industrial settings. 3) Inventions created with AI assistance where AI serves as a tool aiding human inventors in tasks like material discovery or process optimization. 4) Inventions autonomously generated by AI, where human contribution is minimal or absent.
The chapter also explores pressing legal issues, including the recognition of inventorship, the adequacy of disclosure to meet legal standards, and the ethical implications of granting patents in such rapidly evolving contexts.
2. Recognition of Inventorship
The Guidelines explicitly state that inventors must be natural persons, affirming that AI systems cannot be designated as inventors under current laws. This position ensures adherence to existing legal frameworks and clarifies the roles of individuals in the inventive process. The chapter emphasizes that only those who contribute creatively to the core features of an invention can claim inventorship. For example, in cases where AI aids in generating ideas or solutions, the human’s role in directing, supervising, or refining the process is critical for establishing inventorship.
3. Subject Matter Standards
This section explains how AI-related patent claims must satisfy criteria for patentable subject matter. It clarifies that technical solutions leveraging natural laws, such as innovations in data processing methods or improvements in system efficiency, can qualify for protection. However, abstract rules, mathematical methods, or purely intellectual activities without a clear technical contribution are excluded. Detailed examples are provided to guide applicants in framing claims that highlight the technical aspects of AI innovations, such as incorporating algorithms into practical implementations like medical image analysis or autonomous vehicle systems.
4. Adequate Disclosure
The Guidelines emphasize the necessity of clear and complete disclosure in patent applications, particularly given the “black box” nature of many AI technologies. Applicants are required to provide sufficient details about the algorithms, training processes, data inputs, and outputs to enable others in the field to reproduce the invention. The chapter offers practical advice on how to document algorithmic steps, specify data requirements, and describe application scenarios comprehensively. These measures aim to ensure transparency and reproducibility, addressing common concerns about the opaque decision-making processes of AI systems.
5. Assessment of Inventive Step
This chapter provides detailed guidance on assessing the inventive step for AI-related patents. It explains how AI algorithms contribute to the technical solution of a problem, making them eligible for consideration in inventive step evaluations. Examples demonstrate scenarios where integrating AI algorithms with technical features enhances functionality or achieves unexpected results, such as improving energy efficiency in smart grids or optimizing resource allocation in computing systems. The section stresses the importance of showing how AI-specific features contribute to non-obvious technological advancements.
6. Ethical Considerations
The final chapter addresses the ethical implications of AI-related patents, offering principles to navigate concerns such as algorithm bias, data privacy, and public interest. It encourages applicants to ensure compliance with laws and ethical standards in areas like data usage, security, and the societal impact of AI applications. For example, the chapter highlights the importance of designing algorithms that are fair and unbiased while promoting responsible innovation that aligns with societal values. These considerations aim to foster the development of AI technologies that are both legally sound and ethically responsible.
[Comment]
The draft Guidelines issued by CNIPA provide a timely and comprehensive response to the needs of AI-related patent applications. The inclusion of detailed examples and scenario analysis is commendable, offering practical guidance to applicants in navigating examination policies. However, challenges remain in bridging the gap between abstract AI algorithms and concrete technical applications. While the emphasis on natural law-based technical solutions is clear, the subjective interpretation of technical contributions may lead to inconsistencies in examination. Further, while the inclusion of ethical considerations is important, balancing innovation incentives with restrictions will be crucial to avoid hindering technological progress.