Analyzing the Draft Computer-related Inventions (CRI) Guidelines 2025 – Part 4: Sufficiency of Disclosure

Introduction

This article is part 4 of our multi-blog series on analyzing the Draft Computer-Related Inventions (CRI) Guidelines, 2025. Part 1 discusses the new definitions and terms introduced in the Guidelines, Part 2 discusses the judgments referred to in the Guidelines, and Part 3 discusses the novelty and inventive step sections.

On March 25, 2025, the Controller General of Patents, Designs, and Trade Marks (CGPDTM) released Draft Revised Computer Related Inventions (CRI) Guidelines, 2025. The 2017 Guidelines have undergone significant changes to bring more clarity, precision, and legal alignment to the way CRI patents would be handled in India. By integrating judicial rulings into the guidelines, the new (draft) CRI guidelines become not only more technically sound but also legally robust, giving applicants clearer grounds on which, they can base their patent applications.

In this article, we analyze the Sufficiency of Disclosure section of the Draft CRI 2025 Guidelines.

Sufficiency of Disclosure – A Statutory Requirement

In India, the sufficiency of the disclosure requirement is enshrined under Section 10(4) of the Indian Patents Act, 1970 (“Act”) which mandates that every complete specification shall:

Fully and particularly describe the invention and its operation or use and the method by which it is to be performed;
disclose the best method of performing the invention which is known to the applicant and for which he is entitled to claim protection; and
end with a claim or claims defining the scope of the invention for which protection is claimed;
be accompanied by an abstract to provide technical information on the invention.”

According to the Act, the applicant must describe “what” the invention is and “how” it will be implemented. To meet the “what” criteria, the invention must be thoroughly and specifically stated. To meet the “how” condition, the applicant must also provide the best-known technique for implementing the invention. Therefore, the complete specification should completely and specifically describe the invention to comply with the Act’s requirements. It should also enable a person skilled in the art to work on the invention without the patentee’s help or additional excessive experimentation. The description must be unambiguous, clear, correct, and accurate. It must not contain any statements which may mislead the person skilled in the art to whom the specification is addressed.

The “What” Requirement

To meet the “what” requirement, a complete specification should fully and particularly describe the invention. If the patent applications pertain to hardware-based inventions, such as apparatus, systems, or devices, each and every feature of the invention should be described with suitable illustrative drawings. If the invention is a “method”, the necessary sequence of steps should be clearly stated to distinguish the invention from the prior art, using flowcharts and other information required to accomplish the invention, as well as their implementation mechanism. The specification should specify how distinct components work together as well as how they are connected. It should also specify the desired result/output or the intended consequence of the invention, as well as any intermediary components/steps.

The “How” Requirement

To meet the “how” requirement, a complete specification should describe the best method of performing the invention. The optimum method of carrying out and/or using the innovation should be stated with appropriate illustrations. The specification should not only define the invention’s functionality, but it should also specifically and clearly describe how the invention will be implemented.

Sufficiency according to the CRI Guidelines

Owing to the nature of CRIs, particularly those involving AI, ML, or blockchain, it is often difficult to clearly identify and describe the “what” and “how” of a CRI invention in the complete specification. The Draft Guidelines recognize this challenge and throw some light in this direction and on how the CRI inventions can comply with this requirement.

Importantly, the Draft Guidelines refer to the Hon’ble Madras High Court’s decision in Caleb Suresh Motupalli vs. Controller of Patents,¹ which reiterated that vague references to techniques such as “object-oriented design” or “black-box modernization”—without specific implementation instructions—cannot fulfill the sufficiency requirement. The Court held that merely collating prior art concepts without teaching how they are adapted for the invention’s objective constitutes insufficient disclosure.

The Draft Guidelines illustrate how different categories of CRIs can meet the disclosure and enablement requirements through specific examples. Here, we summarize the requirements in brief. For a complete explanation and examples, please see the complete Guidelines.

AI Inventions

AI-based inventions should disclose:

The input-output relationship with clarity on transformation logic.
The nature and characteristics of training data.
Learning algorithms used, training methods, and performance benchmarks.
Any preprocessing methods and how they affect model behavior.
If reinforcement learning is used, the environment-agent dynamics and reward structure must be detailed.
If the technical effect depends on specific training data traits, those traits (not necessarily the raw data) must be described.

Blockchain Inventions

Blockchain-based inventions should explain:

Cryptographic techniques applied.
The structure of blocks and how data is linked or verified.
Consensus protocols used and their role in ensuring system reliability.
Any hardware–network interactions or interfaces.

Machine Learning Techniques

Applications claiming Novel ML techniques must provide a thorough description, including neural network structure, activation functions, network topology, convergence criteria, metadata, and learning procedures. Each algorithm component should be disclosed just to the amount required to achieve the invention’s claimed technological effects, ensuring that a person skilled in the art may accurately recreate the process.

Our Analysis and Conclusion

The Draft Guidelines 2025 significantly improve on the 2017 version in their treatment of sufficiency of disclosure. By offering clear, domain-specific expectations—especially for disruptive technologies like AI/ML and blockchain—they align more closely with global practices. However, challenges remain.

The primary concern is feasibility: how much detail can inventors reasonably disclose without sacrificing trade secrets or overburdening the specification with implementation code? The USPTO’s AI Guidance also recognizes the tension between enablement and trade secrets and permits certain redactions or generalizations as long as they do not impair enablement. Similarly, describing an AI model’s logic can be difficult when the model evolves post-deployment. Therefore, while the Draft Guidelines provide much-needed clarity, formal procedural safeguards for balancing public disclosure with confidential know-how could be addressed better.

Having said the above, the Draft Guidelines do offer a practical roadmap. For applicants, they highlight the need to invest time upfront in drafting specifications that explain the invention in engineering detail, not just functional terms. For examiners, they clarify the boundary between patentable subject matter and abstract ideas.

In sum, the sufficiency of disclosure in CRIs is no longer a matter of reciting system architecture or declaring “AI is used.” However, the success of these Guidelines will hinge not on their articulation but on whether the Patent Office can consistently apply them without drifting into either over-exclusion or over-disclosure—two ends of the same dangerous spectrum.