Patentability of AI Inventions under Patents Act, 1970

Artificial Intelligence (AI) has moved far beyond being a futuristic concept. From healthcare diagnostics and self-driving cars to legal research and financial predictions, AI is actively reshaping industries. With such rapid growth, an important legal question arises: Can AI-related inventions be patented in India under the Patents Act, 1970?

This article explores the scope of patentability of AI inventions under Indian law, analysing statutory provisions, guidelines, and practical challenges. It is written in simple and clear Indian English to help students, lawyers, and innovators understand the subject.

Understanding the Basics: What Makes an Invention Patentable?

The Patents Act, 1970 lays down the conditions for patentability in India. For any invention to qualify, it must meet the following:

1. Novelty – The invention must be new and not disclosed anywhere in the world before the filing date.
2. Inventive Step – It should not be obvious to a person skilled in the relevant field.
3. Industrial Applicability – The invention should be capable of being made or used in an industry.
4. Subject-Matter Eligibility – The invention must not fall under the exclusions listed in Section 3 and Section 4 of the Patents Act.

When it comes to AI, most debates revolve around two aspects:

• Whether AI-based innovations can satisfy the inventive step requirement.
• Whether they fall under exclusions like mathematical methods, algorithms, or computer programs per se under Section 3(k).

Section 3(k): The Biggest Hurdle for AI Patents

Section 3(k) of the Patents Act excludes the following from patentability:

“A mathematical or business method or a computer program per se or algorithms.”

AI systems rely heavily on algorithms, data processing, and mathematical models. As a result, many AI-related inventions are prima facie caught under this exclusion.

However, Indian jurisprudence has taken a practical approach. If an AI-related invention demonstrates a technical effect or technical contribution, it may be considered patentable despite being computer-based.

For example:

• A mere algorithm for data classification may not be patentable.
• But an AI model integrated into a medical device to improve diagnostic accuracy could qualify as it shows a technical application beyond a mathematical method.

Guidelines for Computer-Related Inventions (CRIs)

The Indian Patent Office (IPO) has issued Guidelines for Examination of Computer-Related Inventions (CRIs), which directly impact AI inventions. These guidelines clarify that:

• A computer program “per se” is not patentable.
• If the invention demonstrates novel hardware or a technical improvement achieved through software, it may be considered.
• The emphasis is on the substance of the invention, not just the form in which it is claimed.

For AI, this means:

• Pure algorithms or machine learning models may face rejection.
• AI inventions combined with specific hardware, or demonstrating real-world technical advancement, may succeed.

Examples of Potentially Patentable AI Inventions in India

1. AI in Healthcare Devices – An AI-powered imaging tool that detects tumours with greater accuracy.
2. AI in Agriculture – A sensor-based system using AI for soil analysis and crop prediction.
3. AI in Manufacturing – Robotic arms guided by AI for precision assembly in industries.
4. AI in Transportation – An AI system that reduces fuel consumption by optimising engine performance.

In each case, the AI component is tied to a practical application, making it more likely to be seen as patentable.

Inventorship and AI: Who Owns the Patent?

A growing challenge worldwide is whether an AI system itself can be considered an “inventor.” In India, under the Patents Act, an inventor must be a natural person.

This means:

• AI cannot be named as an inventor.
• Only a human creator, developer, or the entity controlling the AI can apply for a patent.

Thus, even if an AI system autonomously generates a new solution, the patent must be filed in the name of a human or legal entity.

Case Law and International Influence

Though India has not yet witnessed landmark judgments specifically on AI inventions, global trends influence its interpretation.

• In the DABUS cases in the US, UK, and Europe, courts rejected the idea of AI as an inventor.
• India, following a similar principle, requires human inventorship.

Moreover, Indian patent practice often looks at European Patent Office (EPO) standards, which allow AI-related inventions only if they show a technical character.

Challenges in Patentability of AI Inventions

1. Defining Inventive Step: AI systems can “learn” and generate outcomes. Determining whether an AI-generated solution involves an inventive step can be tricky, since AI may mimic human intelligence.
2. Disclosure Requirement: Section 10 of the Patents Act requires the applicant to fully and clearly describe the invention. With AI, explaining the “black box” nature of deep learning models becomes difficult.
3. Rapid Technological Evolution: AI evolves at a faster pace than the patent examination process. By the time a patent is granted, the technology may have advanced, raising concerns of relevance.
4. Overlap with Other IP Regimes: Sometimes, AI-related creations may be better protected under copyright (for source code) or trade secrets (for training data) rather than patents.

Practical Tips for Patenting AI Inventions in India

For innovators and businesses working on AI, the following strategies may improve chances of success:

1. Focus on Technical Application – Highlight how the AI invention improves a technical process or solves a real-world problem.
2. Avoid Claiming Pure Algorithms – Draft claims that link AI methods to hardware or industrial processes.
3. Detailed Specification – Provide clear descriptions of how the AI system works, including architecture, data flow, and practical examples.
4. Hybrid Protection Strategy – Consider using a combination of patents, copyrights, and trade secrets to protect different aspects of AI.

Conclusion

The patentability of AI inventions under the Patents Act, 1970 is both complex and evolving. While Section 3(k) creates a significant hurdle, Indian law allows room for AI innovations that demonstrate technical effect and practical utility. AI itself cannot be an inventor, but human creators and companies can seek protection for AI-based inventions.

As India continues to push towards becoming a global hub for innovation, it is essential to strike the right balance between encouraging AI research and preventing monopolisation of abstract algorithms. 

With careful drafting, strategic planning, and a focus on technical contribution, AI-related patents in India are not just possible but also promising.