- in European Union
The China National Intellectual Property Administration (CNIPA) recently released amended Patent Examination Guidelines that will take effect on January 1, 2026. The amendments cover a wide range of topics, including new requirements for inventor identity, ethical standards for inventions involving artificial intelligence (AI), standards for disclosure and inventiveness of algorithm or data-related inventions, examination of inventions involving bitstreams, patent protection scope for plant varieties and biological breeding, and improvements to invalidation procedures.
A summary of some of the key amendments follows.
I. Strengthened Requirements for Inventor Identity
The amendments to the Patent Examination Guidelines (“amendments”) strengthen the authenticity requirements for inventor identity information and further underscore the duty patent firms have to verify this information. A newly added provision specifies:
The inventor must be an individual (i.e., a natural person). All inventors' identity information must be provided in the request form, and such information must be truthful.
The request form may not list any institution, group, or AI system as an inventor—for example, “XX Research Team” or “AI-XX.”
To prevent false attribution, the amendments explicitly prohibit the inclusion of false inventors. Further, the amendments impose a duty on patent agencies to verify the applicant's identity and contact information indicated in the request form.
These amendments are responsive, in particular, to challenges arising from the development of AI and its implications for inventorship. They reaffirm the fundamental principle that inventors must be natural persons and eliminate the practice of attributing inventorship to research groups or institutions. The changes are consistent with the international consensus on AI inventorship and further align China's patent system with international standards.
II. Examination of Inventions Related to Artificial Intelligence, Big Data, and Similar Technologies
In response to the rapid growth in AI, big data, and algorithm-driven inventions, the amendments establish clearer examination rules for patent applications containing algorithmic or business-rule features, including ethical review, inventiveness assessment, and disclosure requirements.
First, the amendments explicitly state that inventions containing algorithmic or business-rule features will not be granted patent rights if data collection, labeling, rule setting, recommendation decision-making, or other processes involve violations of laws, social morality, or the public interest.
For inventions that include both technical features and algorithmic or business-rule features, the amendments revise and refine the inventiveness evaluation criteria and introduce new examination examples. The amendments emphasize that examiners should focus on the contribution of algorithmic features to the technical solution. If the algorithm and technical components are closely integrated and collectively produce technical effects, they must be evaluated as a whole for inventiveness.
Notably, the disclosure requirements for inventions involving AI models and algorithms have been strengthened. For inventions related to the construction or training of AI models, the description must generally disclose necessary model modules, layers or connections as well as training steps and parameters. For inventions involving the application of an AI model or algorithm in a specific field or scenario, the description must clearly disclose how the model or algorithm is combined with the particular field or scenario, including input/output data settings that demonstrate their internal relationship. This ensures that a person skilled in the art can carry out the invention based on the description.
III. Examination Rules for Inventions Involving Bitstreams
In fields such as streaming media, communication systems, and computer systems, data is typically generated, stored, and transmitted in the form of bitstreams. The amendments introduce new examination provisions specifically addressing inventions involving bitstreams. Under the new rules, a bitstream per se is not a patentable subject matter; however, bitstream-processing technologies derived from a specific video encoding method may be patent-eligible.
The amendments further clarify the drafting requirements for descriptions and claims involving bitstreams. For inventions involving bitstreams generated by a specific video encoding method, the description must clearly and fully disclose the encoding method so that a person skilled in the art can implement it. If the claims seek to protect methods of storing or transmitting the bitstream, or storage media of the bitstreams, the description must include appropriate supporting disclosures.
IV. Optimization of Patent Invalidation Procedures
Iterative attacks on patents based on use of very similar arguments have been a persistent problem in China. While the current rules bar filing repeated invalidation requests based on “the same reasons and evidence”, this has been interpreted very narrowly such that repeated attempts have been permitted even when the “reasons and evidence” change only slightly. This, in turn, has allowed infringers and competitors to tie up patents in never-ending cycles of invalidation proceedings. The amendments now broaden the relevant provision to include “the same or substantially the same reasons and evidence.”
Specifically, after an invalidation decision has been issued for a patent, any new invalidation request based on the same or substantially the same reasons and evidence will not be accepted. Only reasons or evidence that were not considered previously due to time limits or other procedural constraints may qualify as exceptions.
Additionally, the amendments provide that an invalidation request will not be accepted if it is shown that the request does not reflect the true intention of the requester. For example, if evidence suggests that a request was filed at the instruction of another party, or that the requester's identity is false, the request may be rejected. This provision targets malicious invalidation requests or those filed under false identities—such as competitors repeatedly attacking a patent by using third parties' names, or improper agents persuading unrelated persons to file invalidations for illicit gains.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
[View Source]
