Patentability of AI-Assisted Inventions in India: Rethinking the ‘Inventor’ Requirement under Patent Law

Introduction

Artificial Intelligence (AI) has rapidly become a part of scientific and technological research. It is now being used in areas such as drug discovery, medical research, engineering, material science and product development. AI systems can process huge amounts of information, identify patterns and generate possible solutions to technical problems. In some cases, the contribution of AI may go beyond simply assisting a researcher and may become an important part of how the final invention is developed. This development has created a new question for patent law: when AI is involved in creating an invention, who should be recognised as the inventor?

Traditionally, patent law has been based on the assumption that an invention is created by a human being. A person identifies a problem, applies knowledge and skill, develops a solution and seeks patent protection for it. AI challenges this traditional understanding because it can now generate solutions that may not have been directly conceived by the human using the system.
However, it is important to understand that every invention involving AI is not necessarily an invention created by AI. There is a difference between an invention where a human uses AI as a tool and an invention where AI independently generates the technical solution.

This distinction has become particularly important in India after the Guidelines for Examination of Computer Related Inventions (CRIs), 2025, issued by the Office of the Controller General of Patents, Designs and Trade Marks (CGPDTM). The Guidelines specifically distinguish between AI-generated inventions and AI-assisted inventions. They state that AI-generated inventions cannot be patented on the basis that AI itself cannot qualify as the “person” claiming to be the true and first inventor under Section 6 of the Patents Act, 1970. At the same time, AI-assisted inventions are not automatically excluded from patentability merely because AI was used in the inventive process.1

The issue, therefore, is not simply whether AI was used. The more difficult question is what role did AI play and what contribution did the human make to the invention?

Understanding AI-Assisted and AI-Generated Inventions

Before examining the legal position, it is necessary to understand the difference between the two.

AI-Assisted Inventions

An AI-assisted invention is an invention in which AI is used as a tool during the inventive process, but the human remains meaningfully involved in developing the invention.

For example, a scientist may use an AI system to identify possible chemical combinations for treating a particular disease. The AI may generate several possible combinations, but the scientist may then examine those results, select one, modify it, conduct experiments and finally develop a new compound.

In such a situation, AI has assisted the research, but the human has made an important contribution to the final invention.

The important features of an AI-assisted invention are:

The human identifies or works towards solving a technical problem.
AI is used as a tool to generate, analyse or refine possible solutions.
The human evaluates the results produced by AI.
The human may select, modify or improve the AI-generated solution.
The human may conduct experiments or further technical development.
The final invention can therefore be connected to a recognisable human contribution.

Therefore, the mere use of AI does not automatically prevent a person from being recognised as an inventor.

AI-Generated Inventions

An AI-generated invention presents a more difficult situation. Here, the AI system creates the technical solution autonomously or with very limited human involvement. For example, suppose an AI system is provided with general technical information and independently identifies a problem, develops a solution and produces a new technical result without a human selecting or developing the particular solution.

In such a situation, simply owning or operating the AI system may not necessarily make the owner the inventor. The difficult question becomes whether that person actually contributed to the conception of the invention.

The main distinction can therefore be understood through the role of the human in the inventive process:

In an AI-assisted invention, the human remains actively involved in developing the invention, while AI functions as a supporting or assisting tool.
In an AI-generated invention, AI plays a much more independent role in producing the technical solution, while human involvement in the actual conception of the invention may be limited.

In an AI-assisted invention, the human may identify the problem, provide technical inputs, evaluate different AI outputs, select a particular solution and further develop it.
In an AI-generated invention, merely creating, owning or operating the AI system may not be enough to establish inventorship.

The central question is therefore not “Was AI used?”, but “What contribution did the human make to the invention?” This distinction is particularly important under Indian law because the present patent framework requires the inventor to be a legally recognised person. The 2025 CRI Guidelines expressly recognise this distinction.2

The Concept of an “Invention” under Indian Patent Law

The Indian patent system is primarily governed by the Patents Act, 1970. Section 2(1)(j) defines an “invention” as a new product or process involving an inventive step and capable of industrial application.3

Therefore, an invention must satisfy three basic requirements:

1. Novelty – the invention must be news.

2. Inventive step – the invention must involve a technical advancement or economic significance and must not be obvious to a person skilled in the art.
3. Industrial applicability – the invention must be capable of being made or used in industry.

The involvement of AI does not remove these basic requirements. An AI-related invention must still satisfy the ordinary requirements of patentability.

However, AI creates an additional question: who should be identified as the inventor?

The “True and First Inventor” under Indian Patent Law

Section 6 of the Patents Act, 1970 deals with persons entitled to apply for patents. It allows an application to be made by a person claiming to be the true and first inventor, by an assignee of that person, or by the legal representative of a deceased person who was entitled to make the application.

This provision becomes important in the context of AI because Indian patent law connects the right to apply for a patent with the concept of a human inventor.
The requirement is also reflected in the patent application procedure. Form 5, prescribed under the Patents Rules, 2003, is titled “Declaration as to Inventorship” and requires the identification of the true and first inventor or inventors.4

This creates a problem where the invention is claimed to have been independently created by AI.
An AI system cannot presently be treated as the “person” claiming to be the true and first inventor under the Indian patent framework. Therefore, the present Indian position does not allow an AI system itself to be named as the inventor.

The Indian Patent Office’s Approach to AI-Related Inventions

The position became considerably clearer with the CRIs Guidelines, 2025. The Guidelines divide AI-related inventions into two categories:

AI-generated inventions: The Guidelines describe AI-generated inventions as inventions created by AI systems autonomously or with very limited human intervention. According to the Guidelines, such inventions are not patentable because AI cannot be treated as a “person” claiming to be the true and first inventor under Section 6 of the Patents Act, 1970.

AI-assisted inventions: AI-assisted inventions are inventions in which AI is used as a tool during the inventive process.

The Guidelines make it clear that such inventions are not categorically non-patentable merely because AI has been used. They can be considered for patent protection if they satisfy the applicable patentability requirements and demonstrate the required technical effect through their tangible inventive application.6

This is an important distinction. For example, if an engineer uses an AI system to generate different designs but personally evaluates, modifies and develops one of those designs into a working technical product, the use of AI itself does not prevent the resulting invention from being considered for patent protection.

Section 3(k) and AI-Based Inventions

Another important provision is Section 3(k) of the Patents Act, 1970. Section 3(k) excludes from patentability a mathematical or business method, a computer programme per se and algorithms. This provision is particularly relevant to AI because AI systems are based on algorithms, mathematical models and computer programmes.

However, it is important not to confuse an AI-based invention with an invention that is merely an algorithm or computer programme.

A mathematical formula or abstract AI model by itself may fall within the exclusions under Section 3(k). However, when an AI/ML system is applied to produce a real-world technical result, the patentability analysis may be different.

The 2025 CRI Guidelines recognise that abstract mathematical or theoretical AI/ML concepts are not patentable merely because they are described in technological language. At the same time, an AI/ML innovation that transforms an abstract principle into a real-world tangible application may constitute patentable subject matter, subject to the other requirements of patent law.7

Therefore, two separate questions must be considered:

First: Is the subject matter itself patentable under Section 3(k)?

Second: Who is the legally recognised inventor of that subject matter?

A human inventor may exist but the invention may still fail Section 3(k). Similarly, an AI-assisted invention may satisfy the subject-matter requirements but still raise questions about whether the correct human inventor has been identified.

Why the Traditional Inventor Requirement Creates Difficulty

The traditional patent system assumes that the person who develops the inventive concept can be identified. AI makes this assumption more complicated.

Consider a situation where a researcher:

identifies a technical problem
provides relevant information to an AI system
designs the parameters for the AI system
asks the system to generate possible solutions
receives an unexpected technical solution
evaluates that solution
conducts experiments
develops the final product.

The question then becomes: who actually invented the claimed invention?
It may not be sufficient to say that the person who operated the AI system is automatically the inventor. At the same time, it may also be inappropriate to treat the AI system as the inventor under the existing Indian legal framework. The more useful question is whether the human made a meaningful contribution to the conception of the claimed invention. This makes human contribution central to the inventorship analysis.

What Should Count as Human Contribution?

Human contribution can take several forms.

A person may contribute by:

identifying the technical problem
deciding what problem the research should address
designing the research objective
selecting relevant data
deciding how the AI system should be used
developing specific technical instructions or parameters
evaluating different AI-generated solutions
selecting the technically useful result
modifying or refining the AI-generated output
conducting experiments
solving technical difficulties encountered during development
converting an AI-generated possibility into a practical and workable invention.

However, not every interaction with AI should automatically make a person an inventor.
For example, merely entering a broad prompt and receiving an unexpected technical solution may not, by itself, establish that the person conceived the resulting invention. The actual contribution must be examined in relation to the claimed invention.

The DABUS Cases

The international debate over AI inventorship is closely connected with the DABUS patent applications. DABUS is an AI system developed by Dr Stephen Thaler. Patent applications were filed in several countries in which DABUS was identified as the inventor rather than Dr Thaler. The issue eventually reached courts and patent offices in several jurisdictions.

United Kingdom

In Thaler v Comptroller-General of Patents, Designs and Trademarks, the UK Supreme Court considered whether DABUS could be recognised as an inventor. The Court held that an inventor under the UK Patents Act 1977 must be a natural person. It also rejected the argument that Dr Thaler could obtain patent rights merely because he owned the AI system.8 The decision therefore maintained a human-centred approach to inventorship.

Comparative Position in Other Jurisdictions

The approach taken by other countries is useful because it shows how different patent systems are dealing with the same problem.

United States: The United States currently follows a human-inventor approach.

In 2025, the USPTO revised its guidance on AI-assisted inventions and clarified that the same legal standard for determining inventorship applies whether or not AI was used. Only natural persons can be named as inventors. AI systems are treated as tools that may assist human inventors. This is significant because the US approach does not create a separate inventorship test merely because AI was used. Instead, the ordinary principles of inventorship continue to apply.

European Patent Office: The European Patent Office has also maintained that an inventor must be a human being.

In J 8/20, the EPO Legal Board of Appeal rejected an application that identified DABUS as the inventor. The EPO held that the European Patent Convention requires the inventor to be a natural person.10 However, the position became more nuanced with the EPO Board of Appeal decision T 0528/25, decided on 5 February 2026.

The Board stated that, in principle, an inventor can be designated for an invention developed using AI. However, in that particular case, the applicant's statements were considered unclear and inconsistent because he simultaneously identified himself as inventor while stating that DABUS had independently conceived the invention. The appeal was ultimately dismissed.11
The decision is important because it demonstrates that the mere involvement of AI does not automatically prevent an invention from having a human inventor. What matters is whether the legal requirements for designation of a human inventor are properly satisfied.

Australia: Australia also experienced significant developments through the DABUS litigation.

The Federal Court initially accepted the possibility of an AI system being named as an inventor. However, that position was subsequently overturned on appeal, and the requirement of a human inventor was restored. The Australian experience demonstrates that allowing AI to be treated as an inventor would require a significant change in the existing statutory framework rather than merely a broad interpretation of the existing term “inventor.”

South Africa: South Africa has attracted attention because a DABUS-related patent was granted with DABUS identified as the inventor. However, South Africa's patent registration system does not involve the same substantive examination process as many other jurisdictions.

Therefore, the South African development should not automatically be treated as evidence that a broad international consensus has emerged in favour of AI inventorship. The EPO notes that South Africa remains the only jurisdiction that has accepted the designation of an AI system as inventor in the DABUS context.12

The Present Position in India

The present Indian position can therefore be understood through three connected principles.

AI itself cannot presently be treated as the inventor.

Section 6 of the Patents Act refers to a person claiming to be the true and first inventor. The 2025 CRI Guidelines specifically state that AI-generated inventions cannot be treated as patentable on the basis of AI being the inventor because AI cannot be considered the relevant person under Section 6.13

AI-assisted inventions are not automatically excluded.

Where AI is used as a tool but a human makes a meaningful contribution to the inventive process, the invention is not automatically unpatentable merely because AI was involved.14

Ordinary patentability requirements continue to apply.

The invention must still satisfy requirements such as novelty, inventive step, industrial applicability and the other provisions of the Patents Act, including Section 3(k), where relevant.

Therefore, AI does not create a completely separate patent system. Rather, it creates a new question within the existing system: how should human inventorship be determined when AI has contributed to the inventive process?

The Need for Greater Clarity in Indian Law

Although the 2025 CRI Guidelines provide useful clarification, some questions remain open.

For example, Indian law does not provide a detailed statutory test explaining exactly how much human involvement is required before a person can be treated as the inventor of an AI-assisted invention.

This can create uncertainty in cases where AI's contribution is substantial. Suppose an AI system produces ten possible technical solutions. A researcher identifies the most promising one, modifies it, conducts experiments and finally develops the patented product.
In another situation, the researcher may merely provide a general instruction to the AI and receive a complete technical solution without making further changes.

The level of human involvement is clearly different in these two situations. However, the legislation does not currently provide a detailed statutory framework for drawing this line. This is where further guidance could be useful.

Possible Factors for Determining Human Inventorship

India could consider developing clearer guidance based on the actual contribution made by the human.

The following factors could be relevant:

1. Identification of the problem: Whether the human identified the particular technical problem that the invention seeks to solve.
2. Direction of the research: Whether the human gave the AI a specific technical direction rather than simply asking it to independently create something new.

3. Selection of the solution: Whether the human evaluated different AI-generated possibilities and selected the particular solution that became the invention.

4. Modification of AI output: Whether the human made technical modifications or improvements to the AI-generated result.

5. Experimental work: Whether the human conducted experiments to establish that the proposed solution actually works.

6. Technical development: Whether the human solved technical problems or made further developments necessary to convert the AI-generated idea into a workable invention.

7. Connection with the claims: Most importantly, whether the human's contribution is connected to the claimed features of the invention for which patent protection is being sought.

These factors would help distinguish a genuine human contribution from merely operating an AI system.

Importance of Disclosure and Record-Keeping

AI-assisted inventions also raise a practical issue concerning documentation.

Section 10(4) of the Patents Act requires the complete specification to fully and particularly describe the invention and the method by which it is to be performed. In an AI-assisted invention, maintaining proper records of the development process canvtherefore become important.

An inventor or applicant may need to preserve information such as:

the original technical problem
the human researcher's contribution
the role played by the AI system
relevant inputs or parameters
significant AI-generated outputs
experiments conducted after receiving AI outputs
modifications made by the human
the development of the final invention.

Such records may help establish how the invention was actually developed if inventorship is subsequently questioned.

This is especially relevant because AI systems can produce outputs that are difficult to reproduce or explain in the same way as a traditional laboratory experiment.

Should AI Be Recognised as an Inventor?

The question of whether AI should eventually be given legal recognition as an inventor is likely to remain part of the larger policy debate. There are arguments on both sides.

Recognising AI as an inventor could theoretically reflect situations where an AI system genuinely produces a technical solution without meaningful human conception. However, doing so would also raise several practical legal questions.
For example:

Who would own the patent?
Could an AI system transfer patent rights?
Who would be responsible for infringement?
Who would receive royalties?
Could an AI system be a party to an assignment?
Who would be responsible for maintaining the patent?
How would inventorship disputes involving AI be resolved?
Existing patent systems are generally structured around human inventors and legal entities capable of holding and transferring rights.

For this reason, the immediate solution may not necessarily be to give AI the legal status of an inventor. A more practical approach could be to develop clearer rules for determining the human contribution in AI-assisted inventions.

Recent Development: Use of AI in Patent Examination in India

The Indian Patent Office has also begun addressing AI from another perspective: the use of AI by patent examiners themselves.

In August 2026, the CGPDTM published the Guidelines for the Use of Artificial Intelligence in Patent Examination Procedures. These Guidelines address the responsible use of AI during patent examination procedures.15

This development is different from the question of AI inventorship. It concerns the use of AI by patent authorities rather than whether AI can be an inventor.
Nevertheless, it shows that the Indian patent system is increasingly having to address AI at different stages of the patent process. The development also highlights the continuing importance of human oversight in patent examination.

Rethinking the “Inventor” Requirement

The growing use of AI does not necessarily mean that the traditional concept of an inventor has become useless. Instead, it shows that the concept may need to be applied more carefully.
The patent system was developed at a time when the inventive process was assumed to be entirely human. Today, the process may involve humans, AI systems, datasets, simulations and automated experimentation.

The law therefore needs to distinguish between:
A human who genuinely develops an invention with the assistance of AI
A person who merely owns or operates an AI system that independently produces the invention.
Treating both situations in exactly the same way may create uncertainty.
A clearer framework focusing on human contribution to the claimed invention could provide greater certainty without immediately giving AI an independent legal identity.

Conclusion

Artificial Intelligence is changing the way inventions are developed. It can assist researchers in identifying problems, analysing information, generating possible solutions and developing new technical applications.

As AI systems become more capable, the boundary between a human-created invention and an AI-generated invention may become increasingly difficult to identify.

Indian patent law presently maintains a human-centred approach to inventorship. Section 6 of the Patents Act connects the right to apply for a patent with the true and first inventor, while the 2025 CRI Guidelines specifically distinguish between AI-generated and AI-assisted inventions. AI-generated inventions are not treated as patentable where AI itself would have to be regarded as the inventor, whereas AI-assisted inventions are not automatically excluded merely because AI was used as a tool.16

The real difficulty therefore lies in determining when human involvement is sufficient to establish inventorship.

A researcher who uses AI to generate possibilities but then selects, modifies, tests and develops the final invention presents a very different situation from a person who simply owns an AI system that independently produces a technical solution. The law needs to be capable of recognising this difference.

The experience of jurisdictions such as the United States, United Kingdom and European Patent Office shows that most existing patent systems continue to require a human inventor, although they are increasingly developing guidance to deal with inventions created with the assistance of AI.17

For India, the immediate need may therefore not be to replace the human inventor requirement, but to clarify how human contribution should be assessed when AI plays a substantial role in the inventive process. Clearer guidelines on conception, selection, modification, experimentation and technical development could provide greater certainty to inventors and patent applicants.

Ultimately, the challenge for patent law is to recognise the changing reality of technological innovation without losing clarity about who is legally responsible for an invention. AI may increasingly become an important part of the inventive process, but the question of who should receive patent rights will continue to require a clear connection between the claimed invention and a legally recognised inventor.

REFERENCES

[1] Office of the Controller General of Patents, Designs & Trade Marks, Guidelines for Examination of Computer Related Inventions (CRIs) – 2025, para. 5.2.1.

[2] Ibid.

[3] The Patents Act, 1970, § 2(1)(j).

[4] The Patents Rules, 2003, Rule 13(6); Form 5, Declaration as to Inventorship.

[5] CGPDTM, Guidelines for Examination of Computer Related Inventions (CRIs) – 2025, para. 5.2.1.

[6] Ibid.

[7] Ibid., discussion concerning AI/ML/DL inventions and Section 3(k).

[8] Thaler v Comptroller-General of Patents, Designs and Trademarks [2023] UKSC 49. The UK Supreme Court held that an inventor under the Patents Act 1977 must be a natural person and that ownership of DABUS did not itself give Dr Thaler entitlement to the patents.

[9] United States Patent and Trademark Office, Revised Inventorship Guidance for AI-Assisted Inventions (26 November 2025). The USPTO clarified that the same inventorship standard applies regardless of whether AI was used and that only natural persons may be named as inventors

[10] European Patent Office, J 8/20, confirming that an AI system such as DABUS cannot be designated as the inventor under the European Patent Convention.

[11] European Patent Office, Board of Appeal, T 0528/25, 5 February 2026. The Board stated that designation of an inventor is, in principle, possible for inventions developed using AI, but dismissed the appeal because the applicant's designation was unclear and inconsistent with the accompanying statements.

[12]European Patent Office, Inventorship, updated 14 August 2026. The EPO states that, under the EPC, only a human being can be named as an inventor and notes the different treatment of DABUS in South Africa.

[13] CGPDTM, Guidelines for Examination of Computer Related Inventions (CRIs) – 2025, para. 5.2.1.

[14] Ibid.

[15] CGPDTM, Guidelines for the Use of Artificial Intelligence in Patent Examination Procedures, published 7 August 2026.

[16] CGPDTM, Guidelines for Examination of Computer Related Inventions (CRIs) – 2025, para. 5.2.1.

[17] USPTO, Revised Inventorship Guidance for AI-Assisted Inventions (2025); Thaler v Comptroller-General of Patents, Designs and Trademarks [2023] UKSC 49; EPO, Inventorship.