AI Generated Inventions And Its Patentability Under Patent Law



For the first time in the history of intellectual property, the South African patent office has granted a patent to an AI. You did read it correctly—a patent whose creator is artificial intelligence. The name of this AI is "DABUS" (Device for Autonomous Bootstrapping of Unified Sentience). It was created by Dr Stephan Thaler and is used by Professor Abbott and his team at the University of Surrey. It is a "creative machine" that can produce ideas without the need for human input. DABUS has received training in the creation of novel concepts; most recently, DABUS independently developed the two inventions that are filing for patents. Dr Thaler, who holds the patent, filed the registration in 2019, but DABUS was listed as the inventor. These patent applications have also been submitted in several other nations, and the authorities have responded to them in a variety of ways. Primarily focusing on India's take, it seems pretty difficult to consider an AI invention patentable under the Indian Patent Act.

patent act[Image Sources : Shutterstock]

Since DABUS is not recognized as a person under Sections 2 and 6 of the Patents Act, 1970, the Controller General of Patents in India stated objections in the Examination Report of Thaler's Indian patent application, claiming that the application could not pass the formal and technical examination. This was supported by numerous judicial precedents. For example, in the case of V.B. Mohammed Ibrahim v. Alfred Schafranek, the Court decided that neither a financing partner nor a corporation can be the sole applicant as an inventor and held, among other things, that only a natural person who truly contributes their skill or expertise towards the innovation can claim inventorship under the law. As a result, it is not possible to recognize AI as patent holders under India's current statutory framework.


Talking about Artificial Intelligence is the ability of a computer or other machine to mimic the skills of the human mind, which frequently develops these abilities from prior experiences to comprehend and respond to language, choices, and challenges. It is the technology we've developed that enables computers and other machines to function intelligently. It’s a machine that takes the place of labourers to produce results more quickly and effectively. Others still see it as "a system" that can accurately understand external data, learn from it, and then apply what it has learned to accomplish particular objectives and activities through adaptability .

We have laws that safeguard a person's tangible ideas as intellectual property when it comes to intelligence. The Indian Patents Act, passed in 1970, lays out the guidelines for granting intellectual property rights to those who invent novel and ground-breaking innovations. With the development of technology, AI is now able to create original concepts that are tangible and worthy of being protected by patents. The question is, however, whether or not AI is recognised under our current legal system as an inventor, and if so, what benefits and drawbacks there may be. We'll talk about it from the standpoint of Indian patent law .


A person may apply for a patent under section 6 of the Indian Patent Act, as may their assignee or the legal representative of any deceased. The government is included under the Indian Patent Act's definition of "person." Therefore, the patent may be filed by either a natural person or a government. There is no mention of the machine being listed as an inventor on a patent, though. As a result, it is unclear who is entitled to intellectual property rights in any idea or work that the AI is responsible for. The parties will need to rely on strong contracts with precise ownership clauses up until the law is changed to include machines as inventors. If there is a disagreement, the parties agreement and the facts of the case will be taken into consideration by the courts in making their decision. There aren't any court rulings concerning who owns the intellectual property rights to AI inventions as of yet.

The first is the hotly contested question of whether or not the machine may legitimately claim to be an inventor. The term "inventor" is not defined anywhere in the act specifically; instead, it allows for the filing of patent applications by, among others, the person(s) claiming to be the true and original inventor and the assignee of such "person(s)". The phrase "true and first inventor"[ Section 2(1)(y)] and the definition of the word "person" [Section 2(1)(s)] do not help the cause. Therefore, the Act cannot be used to provide a solution to this problem. Shining Industries vs. Sri Krishna Industries and V.B. Mohammed Ibrahim vs. Alfred Schafranek & Ors. provide some clarification, but only when establishing whether a financial contribution or legal entity can claim inventorship.


A business technique or mathematical, an algorithm or a computer programme, are not considered to be inventions, according to Section 3(k) of the Indian Patents Act. It places limitations on the general patentability of computer programmes.

The Delhi High Court (DHC) ruled in a writ petition challenging the IPAB's decision to deny Ferid Allani a patent on a "method and device for accessing information sources and services on the web" that the IPAB's decision will be based on "technical effect" and "technical contribution." Therefore, the technologies will qualify for consideration for the patent grant if they can solve a technological problem through technical means.

Because they are typically seen as having a mathematical nature and not patentable, computational models and algorithms . This has become the main reason why many Indians who are unable to patent their research/inventions in India are applying in the USA. However, it is often forgotten that such models can add to the technical nature of an invention by generating a technical effect that furthers a technical objective, by i) being applied to a field of the invention; and/or ii) being tailored to a particular technical implementation. Therefore, depending on how it is claimed, a mathematical model or computational or AI algorithm should not be disqualified as being ineligible if it satisfies these and other criteria.

In conclusion, the procedure of patenting an AI is difficult. Although many nations and organisations have provided detailed guidance on AI patenting, many more have not yet done so. As we evaluated various aspects of patentability across various countries and organisations, we found that the majority of them had placed a strong emphasis on the development of the machine learning-based inventions and how they are technically and practically superior to previous ones.

The world is quickly moving toward the future and becoming more and more reliant on AI-generated inventions. A more sophisticated and straightforward method of assigning patent rights to an invention created by an artificially intelligent system is needed, given the significance of AI-generated breakthroughs.

Author: Aditi Bhushan, A Student of BALLB from KIIT School of Law, Bhubaneshwar, in case of any query, contact us at Global Patent Filing or write back us via email at

Get In Touch