Software Patents in India and Section 3(k): What Is Actually Patentable

India does not grant patents on computer programs "per se". Section 3(k) of the Patents Act excludes software, algorithms and business methods as such. An invention showing a genuine technical effect or technical contribution may still qualify, whether or not it involves new hardware, but the threshold is high and outcomes vary.

If you build software or an AI product and are weighing patent protection in India, the first thing to understand is a specific statutory exclusion. India treats pure software differently from a machine or a chemical process, and the difference shapes what you can realistically claim.

What section 3(k) actually says

Section 3(k) of the Patents Act 1970 lists, among things that are not inventions, "a mathematical or business method or a computer programme per se or algorithms". The phrase that does the work is "per se". A computer program considered on its own, an abstract algorithm, or a method of doing business is not patentable subject matter in India. This mirrors a broader international pattern where software as such sits outside the patent system, though the exact boundary differs by country. For the wider picture of how software and AI intersect with IP across jurisdictions, see our overview of AI and IP.

The practical effect is that you cannot obtain a patent simply by describing a clever piece of code or a novel business process implemented on a generic computer. A claim framed as "a program that does X" or "a method of trading that does Y" tends to fall squarely within the exclusion.

The technical-effect route and the CRI guidelines

The "per se" wording leaves a narrower door open. Where a claimed invention delivers a genuine technical effect or a technical contribution that goes beyond the ordinary running of a program on a computer, it may be considered patentable. The touchstone in current guidance and case law is that technical contribution, assessed on what the invention actually does, not the form the claim takes. A link to novel hardware can help evidence a technical effect, but it is not a legal requirement, and a software-only invention is not excluded merely because it lacks new hardware.

The Indian Patent Office has issued Guidelines for Examination of Computer Related Inventions (commonly called the CRI guidelines) to address exactly this question. They have been revised over time, the current operative version being the later revision that moved the analysis onto technical contribution, so the version you rely on matters. As a general orientation, examiners look for a real technical contribution and are wary of claims that recite a computer only as a vehicle for an otherwise excluded idea. Because the guidance and the case law around it continue to evolve, treat any summary, including this one, as a starting point and confirm the present position with the Indian IP Office (the CGPDTM) or a qualified Indian patent attorney before you rely on it.

Be conservative in your own assessment. A pure algorithm, a mathematical method, or a business method dressed up in technical language is not patentable in India, and framing alone will not rescue it. What strengthens a claim is a demonstrable technical effect or a real improvement to how a device or system functions; a hardware element can be one useful indicator of that, though it is not decisive on its own. This is a fact-specific analysis, and drafting the claims well makes a material difference to the outcome.

For most software and AI businesses, patents are not the primary protection, and that is not a weakness. Source code and object code are protected as literary works under Indian copyright law from the moment they are created, with no registration required, although voluntary registration is available and can help with evidence. Copyright protects the expression of your code, not the underlying idea or function, but for many products the code itself is the asset worth guarding.

Alongside copyright, trade secret protection carries a great deal of the load for software and AI. Model weights, training data, architecture choices, internal algorithms and know-how can be kept confidential and protected through well-drafted employment terms, non-disclosure agreements and access controls. India protects confidential information through contract and equity rather than a dedicated trade secrets statute, so the strength of your protection depends heavily on the agreements and internal controls you put in place. For many teams a layered approach, copyright over the code plus disciplined trade secret hygiene, gives more durable protection than chasing a patent that section 3(k) may exclude.

Where a patent still makes sense

None of this rules patents out. If your invention genuinely improves how hardware operates, or solves a technical problem in a technical way, a patent may add real value and should be assessed properly. The right sequence is usually to get an experienced Indian patent attorney to review the specific invention and advise on claim strategy before you spend on filing. You can start from our broader India patents section to orient yourself on the wider framework.

On timelines, fees and procedural steps, these change and vary by case, so we have deliberately avoided quoting fixed figures. Official fees apply; confirm the current amount and the up-to-date procedure with the Indian IP Office (the CGPDTM) or local counsel.

A note on how to use this

IPEnvoy is not a law firm and does not provide legal advice; this is general information only. The line between an excluded computer program "per se" and a patentable technical contribution is genuinely difficult, and it turns on the details of your invention and on current guidance, so confirm the position with the Indian IP Office (the CGPDTM) and a qualified local IP professional before acting. If it would help, IPEnvoy can connect you with vetted IP experts in India who handle software and AI matters, so you get advice grounded in the current practice rather than a general summary.

Related

Author: Steffen Hoyemsvoll

Reviewers: pending review