CNIPA's 2025 Amendments to the Patent Examination Guidelines: Key Interpretations and Implications for Applicants (Effective January 1, 2026)

Source: Official Interpretation by CNIPA (December 4, 2025).   https://www.cnipa.gov.cn/art/2025/12/4/art_66_202935.html

Introduction

To implement the important instructions of General Secretary Xi Jinping on intellectual property work and the decisions of the CPC Central Committee and the State Council, and to respond to new requirements for patent examination amid rapid development in emerging fields such as artificial intelligence (AI), the China National Intellectual Property Administration (CNIPA) initiated revisions to the Guidelines for Patent Examination in January 2025.

After public consultation on the draft from April 30 to June 15, 2025, incorporating extensive feedback through written opinions, symposiums, and field research, the final Decision (Order No. 84) was promulgated on November 10, 2025, and became effective on January 1, 2026.

This revision systematically refines examination standards, particularly for AI-related inventions, while optimizing procedural rules to enhance examination quality, efficiency, and innovation incentives.

I.Major Focus: Further Improvements to AI-Related Patent Examination Standards

This amendment primarily targets Part II, Chapter 9, Section 6 (“Examination of Invention Patent Applications Containing Algorithmic Features or Business Rules and Methods Features”), with key enhancements in four aspects:

(I) Clarification of Patentable Subject Matter Boundaries

u Reinforces exclusions under Article 25(1)(ii) of the Patent Law for applications dominated by algorithms or business rules without technical contribution.

u Adjusts numbering and examples for clarity (e.g., previous Example 1 renumbered to Example 3).

(II) Enhanced Inventive Step Assessment

u Emphasizes holistic consideration of technical features interacting with algorithmic/business features.

u Adds new Examples 18 and 19 to illustrate inventive step in AI scenarios, such as model adaptations showing verifiable improvements in accuracy, robustness, or efficiency.

u Non-technical adaptations (e.g., simple data swaps without synergistic technical effect) are likely deemed obvious.

(III) Sufficiency of Disclosure Requirements

u Strengthens rules for describing AI model structures, training processes, and data sets, especially for image recognition, natural language processing, and predictive algorithms.

u Requires clear, reproducible technical details to enable skilled persons to carry out the invention.

(IV) Ethical and Public Interest Considerations

u Introduces explicit rejection grounds for inventions violating ethics or public order (e.g., autonomous vehicle emergency decision models discriminating based on gender/age, conflicting with equal value of human life).

u This is the first systematic clarification of ethical boundaries in AI patent examination.

These changes build on the 2024 Guidance for AI-Related Invention Patent Applications (Trial) and address practical challenges in AI, big data, and related fields, providing clearer guidance for applicants and examiners to promote high-quality innovation.

II. Optimization of Inventive Step Examination Standards

The revision refines Part II, Chapter 4, Section 6 (“Examination of Inventiveness”):

u Emphasizes evaluating the entire technical solution from the perspective of a person skilled in the art.

u Requires applicants to articulate the substantive contribution in claims and specification.

u Discourages reliance on non-technical features for inventiveness; only features contributing to solving technical problems are considered.

This aims to ensure granted patents reflect genuine technological advancement.

III. Implementation of On-Demand Examination Principle

u New Section 8.3 added to Part V, Chapter 7 (“Rapid Examination”): Applications pre-qualified by national or local IP protection centers/fast-track centers may qualify for expedited substantive examination.

u Reinforces the “as fast as needed, as delayed as appropriate” approach, expanding channels for priority, PPH, centralized, and deferred examination.

IV. Other Notable Procedural and Practical Adjustments

u Inventor Requirements: Inventor must be a natural person; request forms must include authentic identity information for all inventors (no groups, collectives, or AI names allowed).

u Dual Filing (Invention + Utility Model): Applicants must abandon the utility model to obtain invention patent grant; no longer allowed to amend claims for different scopes.

u Refund Rules: Limits automatic refunds in certain scenarios (e.g., post-termination/invalidation year fees).

u Plant Varieties & Biological Breeding: Clarifies definitions and exclusions (e.g., artificially improved wild plants may be patentable if industrially applicable).

u Bitstream Inventions: Explicit examination rules for video/audio encoding/decoding bitstreams, requiring technical context disclosure.

Implications for Foreign and Domestic

Applicants These amendments better accommodate emerging technologies like AI, streaming media, and biotechnology, while raising the bar for disclosure, inventiveness, and ethical compliance. Foreign applicants should:

u Pay closer attention to AI ethical/public interest issues in drafting.

u Ensure algorithmic features demonstrate clear technical synergy.

u Prepare more precise specifications for model/training details.

u Consider on-demand examination channels for faster grants.

For tailored advice on filing strategies under the new Guidelines, contact our team of qualified Chinese Patent Attorneys. We track CNIPA updates closely to help global innovators secure robust protection in China.