The Indian Patent Office has also granted a patent for IIT Kanpur’s music training system, which detects performance errors, generates assessment reports, and delivers auditory feedback.[2] Meanwhile, startup Brenin’s patent-pending invention involves an AI-driven digital human technology that leverages psychological insights from real-time conversations to improve employee efficiency, productivity, and well-being.[3]

These are just a few examples highlighted in NASSCOM’s April 2025 report, which indicates that over 86,000 AI patent applications were filed in India between 2010 and 2025, accounting for more than 25% of all technology patent filings. However, India’s share in granted AI patents remains modest at 0.37%, compared to China’s 69.7% and the USA’s 14.2%.

The rapid adoption of digital technologies underscores the need for a regulatory framework for patent examination that can keep pace with innovation. While examining patent applications in cutting-edge fields, it is necessary to consider how innovations go beyond traditional software and algorithms to deliver technical solutions. Patent regimes face particular challenges in processing applications related to Computer Related Inventions (CRIs) and emerging technologies such as AI, machine learning (ML), deep learning (DL), blockchain, and quantum computing.

To foster consistency in the examination of CRI applications, the Office of the Controller General of Patents, Designs & Trade Marks (CGPDTM) released revised guidelines on July 29, 2025.

CRI Guidelines: A Look at Key Features 

Key features of the new guidelines include:

• Effective Date: The guidelines are applicable effective immediately.
• Enforceability: In case of conflict with the Patents Act, 1970, or the Patents Rules, 2003, the latter prevails.
• CRI Applications: The requirements of novelty (including the seven-step approach), inventive step, industrial applicability, and sufficiency of disclosures are elaborated upon. Moreover, the guidelines explain how to assess patentability under the four limbs of Section 3(k): mathematical methods, business methods, computer programmes per se, and algorithms.
• Examination of Inventions Related to AI, ML, etc.: The guidelines highlight the sufficiency of disclosure requirements and aspects that may take the invention out of the purview of exclusion under Section 3(k). As to AI-related inventions, while AI-generated inventions are not patentable, AI-assisted inventions are not categorically excluded under Section 3(k), provided they meet patentability criteria and demonstrate technical effect through their tangible inventive applications.

Does this overhaul mean that the onus now shifts entirely to the patent office and dissuades inventor participation? The answer is far from a simple ‘YES’. On the contrary, the guidelines outline a list of responsibilities for the inventors to ensure a hassle-free patenting process.

Takeaways for Inventors

Here are the takeaways for inventors:

• The necessity for the inventor to articulate the technical problem that the invention addresses and highlight the technical solution to address the problem, a technical contribution that goes beyond the mere functioning of the computer, needs to be brought out.
• The demonstration of the technical contribution or technical effect may be done by referring to some of the generic and non-limiting examples disclosed in the guidelines that explicate the enhancements, notably, increased speed of processing, reduced hard-disk or memory access time, better control of robotic arm, improvement in reception/transmission of radio signal, security enhancement in computer systems/networks, etc.
• A categorical need by the inventors to be ‘detailed but succinct’ in their submissions. The ‘what’ and ‘how’ aspects of the sufficiency standards need to be met, as there is a quantum leap in the examination procedure involving AI, blockchain, IoT, quantum computing and machine learning, and other ultramodern domains.
• For AI systems, the detailed description should aim to:
  • Clarify the logic of transforming the input into the output.
  • Mention of the correlation between the input and output data for a trained AI model.
  • Elucidate the steps and functions related to pre-processing, if the disclosure talks about data pre-processing.
• The inventors need to elaborate on the details related to the structure of the neural network, activation functions, and learning mechanisms, among others, to avoid ambiguities in the examination of the machine learning-related disclosures.
• Since the guidelines also address the subject matter of blockchain-related inventions, the disclosure must comprehensively highlight the details of the cryptographic techniques, data structures, and the consensus mechanism employed.
• Being prepared for the objections received from the examiners for patents falling under non-statutory categories (especially under Section 3(k)).
• The need to refrain from portraying mathematical or business methods, computer programs per se and algorithms as potential inventions. Instead, the efforts should be redirected to the technical problems they address.

Concluding Thoughts

The guidelines appear to have made a sincere attempt to draw parallels from other global jurisdictional examination procedures, such as those used in the United States Patent and Trademark Office (USPTO) and the European Patent Office (EPO). However, the guidelines are adapted to the Indian context, precluding strict exclusions of the harsh Alice/Mayo test used in the USPTO, simultaneously introducing the technical character test of the EPO, ensuring diligence in the examination procedure.

While the guidelines rely heavily on excerpts from judgments, more explicit inferences and a more reader-friendly format could improve usability. Using clearer, illustrative examples in the context of AI-related inventions might add further objectivity to the examination process. Nonetheless, with practical examples and flowcharts, the 2025 guidelines remain a useful reference for navigating CRI examinations.