Yes, if the invention is more than the program itself. Section 3(k) of the Patents Act excludes a computer programme per se, mathematical and business methods, and algorithms. But a software-implemented invention that solves a technical problem and produces a technical effect, such as making a device, network or system work faster, more securely or more efficiently, can be patented. The Patent Office's revised Guidelines for Computer Related Inventions (July 2025) set out this technical effect test, with a dedicated chapter on AI, blockchain and quantum computing.
Founders often hear that "software can't be patented in India". That is only half true. Code by itself is excluded, but the Indian Patent Office grants patents on software-based inventions regularly, from network security and data compression to medical imaging and AI applied to industrial problems. What matters is what your claims protect and how the application explains the technical result.
What Section 3(k) says
Section 3 of the Patents Act, 1970 lists what is not an invention. Clause (k) excludes: "a mathematical or business method or a computer programme per se or algorithms". The key words are "per se": a program as such cannot be patented, but an invention that uses a program to achieve a technical result is not automatically excluded.
How courts and the Patent Office read "per se"
In Ferid Allani v. Union of India (Delhi High Court, 2019), the court held that an invention should not be refused just because it is implemented through software: if it shows a technical effect or technical contribution, it can be patented. The Delhi and Madras High Courts have applied this approach in several later cases involving software and computer-related inventions.
In July 2025, the Office of the Controller General of Patents issued revised Guidelines for Examination of Computer Related Inventions (CRIs), 2025. They give examiners a step-by-step framework for Section 3(k) with worked examples, a chapter on case law, and, for the first time, a dedicated chapter on artificial intelligence, machine learning, blockchain and quantum computing.
What usually gets granted and what doesn't
| Usually patentable | Usually excluded |
|---|---|
| A method that reduces memory use or processing time of a device | A new pricing, discount or loyalty scheme run on a computer |
| A more secure authentication or encryption process for a network | A mathematical formula or algorithm claimed on its own |
| Software that controls hardware: IoT devices, robots, vehicles, machines | An app that only automates a manual business process on a standard computer |
| Image or signal processing that improves the output of a scanner, camera or sensor | Presenting information or a user interface layout as such |
| An AI model applied to a technical problem, with the architecture and training disclosed | An AI model claimed in abstract terms, with no technical application or detail |
Business methods are the hardest category. If the real contribution is commercial (better pricing, a new way of selling, a payment arrangement), the exclusion is very difficult to overcome however much software is involved.
AI and machine learning inventions
The 2025 guidelines expect AI applications to be fully disclosed, so a person skilled in the field could reproduce the result. In practice that means describing the type and source of training data, the model architecture (for example, the layers of a neural network), the training method, and the technical problem the model solves. A claim to "using machine learning to predict X" with no technical detail is likely to be refused. See our AI and machine learning patent guide.
How to draft a software patent that gets granted
- Start from the technical problem. Describe what was technically wrong before (slow, insecure, inaccurate, resource-heavy), not the business goal.
- Show the technical effect with evidence: reduced latency, lower memory or power use, fewer errors, better security. Test data helps.
- Describe how the software works with the hardware: processors, memory, sensors, networks and devices, with flowcharts and system diagrams.
- Claim the method and the system, and consider a computer-readable medium claim; each must carry the technical features, not just the business steps.
- Avoid business language in the claims ("customer", "price", "profit") where a technical description is accurate.
- Search first. Software is a crowded field; a patent search shows what is new before you spend on drafting.
If you still receive a Section 3(k) objection, it can often be overcome in the reply. See how to reply to a patent FER.
If your software is not patentable
- Copyright protects your source code as a literary work automatically; registering it gives you evidence of ownership. See copyright registration.
- Trade secrets protect algorithms and know-how you keep confidential, backed by NDAs and employment agreements.
- Trademarks protect your product name and logo, often the most valuable asset of a software business. See trademark registration.
Getting it right
Whether a software invention gets a patent in India is decided mostly at the drafting stage. At CorpBridge, software and AI patent applications are drafted only by registered patent agents with experience in computer-related inventions. Share your idea in confidence for a free first review of patentability.
Frequently asked questions
Can software be patented in India?
A computer programme per se cannot, under Section 3(k). A software-implemented invention can be patented if it solves a technical problem and produces a technical effect, such as improving how a device, network or system works.
What is Section 3(k) of the Patents Act?
Section 3(k) says that a mathematical or business method, a computer programme per se, or algorithms are not inventions, so they cannot be patented in India.
What are the CRI Guidelines 2025?
The revised Guidelines for Examination of Computer Related Inventions, issued by the Office of the Controller General of Patents in July 2025. They explain how examiners apply Section 3(k), with examples and a dedicated chapter on AI, machine learning, blockchain and quantum computing.
Can an AI invention be patented in India?
Yes, if it applies AI to solve a technical problem and the application discloses enough detail, such as the training data, model architecture and training method, for a skilled person to reproduce it. Abstract AI models or algorithms on their own are excluded.
Can a mobile app be patented in India?
The app's code cannot be patented as such, but a new technical feature it implements, for example a more efficient way of syncing data or securing transactions on the device, may be. The app's code is protected by copyright and its name by a trademark.
Can business methods be patented in India?
No. Business methods are excluded under Section 3(k), and the exclusion is difficult to overcome even when the method is implemented through software.
This guide is general information, not legal or tax advice. Rules and thresholds change; speak to an expert about your situation.