How India’s new guidelines are making sense of patents in the age of AI

The country’s computer related inventions guidelines are the most ambitious so far, providing clarity on patenting computer-related inventions and possibly representing a watershed moment for startups, says Pankaj Soni of Remfry & Sagar.

India’s patent regime has moved a (giant) leap forward. In February 2025, the Indian Patent Office issued its Revised Guidelines for Examination of Computer Related Inventions (CRIs), something many in the technology and innovation sectors had been eagerly anticipating.

Why? Because for years now, engineers working on leading-edge technologies such as artificial intelligence (AI), blockchain, IoT, cloud, and quantum computing have grappled with one huge question: what’s really patentable in India, and what isn’t? Time and again, clarity appeared to emerge, only to recede again into blurred interpretations during examination.

The new guidelines do not resolve every issue, but they do introduce much more clarity to an area that has long been marked by uncertainty. Henceforth, if your invention addresses an actual technical issue and you can explain it clearly, India’s patent regime is ready to listen.

A decade in the making

The journey began back in 2013, when the CRI guidelines were originally published.

There have been several revisions since then to try to make examinations more uniform between different branches of the Indian Patent Office and to keep up with courts’ interpretation of the law.

This year’s revised guidelines are the most ambitious so far. They draw directly from precedent-setting court decisions, add formal tests for difficult categories, and, for the first time, address AI and quantum technologies (though it may be argued that AI deserves its own set of guidelines).

Anchored in court precedent

At the heart of the debate lies Section 3(k) of the Patents Act, which excludes from patentability “mathematical methods, business methods, computer programmes per se, and algorithms”.

These excluded categories overlap with current tech innovations—all in a general way can be anchored to one of the exclusions and, by default, become non-eligible subject matter.

To address this, the Patent Office has moved closer to High Court rulings such as Ferid Allani (2019), Microsoft (2023), Ab Initio Technology (2024), and Blackberry (2024), which made it clear that what is important in assessing patentability is whether an invention provides a technical contribution over and above abstract ideas.

Through integrating these judgments directly into the guidelines, examiners now have more defined perimeters, and innovators understand where they stand.

Breaking down the tests

Perhaps the most useful aspect of the 2025 guidelines is the provision of breakdown tests for frequent problem areas.

• Mathematical methods: If your invention is simply equations, it will not get through. But if taken as a whole it serves some larger technical function, and the equations are applied to real-world systems such as minimising noise in a communication network or encrypting messages, it could qualify.
• Business methods: Strategies dressed up as software do not count. But if your method solves a technical problem (say, improving data processing efficiency), it may be patentable.
• Algorithms: Not all algorithms are barred. If the claimed invention provides enabling details and demonstrates a technical solution to a real-world problem, it deserves examination on merit.
• Computer programmes per se: This has always been a sticking point. A new four-step approach requires examiners to look at the invention as a whole: identify the problem; identify the technical solution; decide if the effect is more than mere incidental impact, and then decide on patentability. Examples serve to illustrate the distinction between unpatentable software and a valid technical innovation.

The guidelines also emphasise that the Patent Office should follow the “Seven Stambh” (seven-step) test from Ericsson v Lava (2024) as the benchmark for assessing novelty.

The steps involve: understanding the claims; identifying the relevant prior art; analysing the prior art; determining explicit and implicit disclosures; assessing material differences in full context; verifying novelty in view of scope and combination; and documenting the analysis.

For inventive step analysis, the focus must be on showing evident technical contribution, rather than relying on incremental innovation. Applications need to detail how the invention operates and vague descriptions or efforts to disguise outdated technology will not be accepted.

This is also critical for means-plus-function claims because the guidelines explain that employing “means” language should be supported with definite descriptions of the hardware, firmware, or software that performs the function.

With respect to AI, machine learning, deep learning, blockchain, and quantum computing technologies, the message of the Patent Office is that the applicant must demonstrate how the invention functions in reality.

If the invention gives a clear technical solution by technical means, then it merits serious consideration. Each technology segment is accompanied by examples, signalling the Patent Office’s intent to bring greater predictability to examination.

Yet examples can be a double-edged sword because what seems helpful today may serve a narrow interpretation tomorrow since technology is evolving continuously.

The reliance on examples also risks rigid interpretation by examiners (which they have a past history of), and whether this approach leads to lasting clarity or proves limiting is something only time will tell.

What this means for startups

For India’s dynamic startup culture, particularly those companies developing AI or blockchain-related innovation, the new guidelines might be a watershed moment.

Before, most young companies steered clear of filing patents entirely, fearing capricious rejections and redundant filing and prosecution fees. Now, with the tests, judicial anchoring, and real-life examples, startups possess a better basis on which to stand their ground in defence of their filings.

This does not imply that all ideas will pass through. But it does suggest that innovations that are technically deep and relevant to the real world are more likely to see a patent issued in India.

The new rules also hopefully cut down uncertainty for investors. Patent portfolios tend to be central to valuations, and more transparent rules can make India a more appealing market for IP-based startups.

Looking ahead

The 2025 CRI Guidelines try to strike a balance between statutory patent policy and innovation incentives. To this end, they make things simpler for examiners by providing clear frameworks, and they provide innovators with a more certain path through India’s patent system.

Challenges are still likely to arise, but that is a by-product of any patent system in the world. For now, this is a pro-innovation signal from the Patent Office, and one that aims to position India alongside the realities of 21st-century technology.

Pankaj Soni is partner and patent chair at Remfry & Sagar. He can be contacted at: pankaj.soni@remfry.com