-Harshal Chhabra* and Arihant Sethia**
Abstract
This article examines the intricate relationship between artificial intelligence (AI) and intellectual property (IP) rights, particularly focusing on the patentability of AI-generated inventions under Indian law. As AI becomes more prevalent in innovation, the question of patent ownership and the criteria for patentability of AI-generated inventions becomes increasingly complex and consequential.
The current Indian patent law does not explicitly address the patentability of AI-generated inventions, leading to ambiguity and potential legal disputes. This article argues for reforms in the Indian patent law to provide clarity and guidance on the patentability of AI-generated inventions.
Specifically, it proposes the adoption of an “Innovation Oversight” approach, which involves assessing the inventive contribution of AI in the context of the overall inventive concept. This approach aims to strike a balance between incentivising AI innovation and ensuring fair access to AI-generated inventions. By advocating for reforms that address the unique challenges posed by AI in the context of patent law, this article seeks to contribute to the ongoing discourse on AI, IP rights, and innovation in India.
Introduction
There is rapid transformation in today’s world surrounded by artificial technology that has exceeded expectations of almost everyone across the globe. This disruption has changed the technology sector and given rise to significant legal challenges in almost every country in the world. The Indian Patent Act 1970, surged way before the AI revolution and, therefore, require amendments to effectively address the technical and legal problems that arise from these disruptive technologies in the economy.
AI technologies’ immense potential for driving economic growth and innovation qualifies them for patent protection. However, identifying the rightful patent holder for AI-generated inventions poses a novel challenge. Should the AI itself be granted a patent, acknowledging its independent functionality? Or should the individual overseeing the AI’s operations and decisions be considered the legitimate owner? Alternatively, given their role in developing and refining the AI system, should the inventor be recognised as the proprietor? These questions underscore the urgent need for a comprehensive AI technology patent framework.
The authors propose an “Innovation Oversight” approach to address this challenge, which emphasises the role of human oversight and decision-making in the AI invention process. The approach emphasizes the role of the innovator in developing the invention and securing its patent. This approach suggests that patents should be granted to individuals or entities that exercise intellectual control over the AI invention, aligning with the fundamental principles of patent law, including the requirement of an inventive step, which entails a new way of doing something or offering a new technical solution to a problem. The most crucial part of this approach is that it does not require a drastic change in the Patents Act, 1970 in order to inculcate the patentability of AI technologies in India.
By adopting this approach, India can establish a robust framework for guiding the patenting of AI-generated inventions in the country in an effective manner. The approach establishes a balance between AI development and potential abuses of the patent system, ultimately contributing to the development of AI-generated inventions in the country.
The article first discusses the necessity of granting patents to AI technology to incentivise innovation. It then evaluates India’s current legal position regarding AI inventions’ patentability. The article further explores the question of patent ownership for AI-generated inventions, considering whether it should be granted to the AI technology, the operator, or the developer. Finally, it introduces the “Innovation Oversight” approach as a solution, emphasising human control and decision-making in the AI invention process to determine patent ownership.
What is the need to grant patents to AI-generated inventions?
The ultimate aim of patents is to give immunity to products which are non-obvious to other skilled players. The basic element for patenting a product is non-obviousness and addition to already existing products. AI alone cannot introduce non-obviousness to an existing product, and therefore, it cannot independently qualify for a patent. Only a skilled individual can add the necessary novelty to a product, making it suitable for future industrial application. That is why the role of AI will be limited to increasing the efficiency in the development of new products and not to producing a new product itself. Due to this, AI technologies must fall in the four corners of the Patent Act.
Drawing parallels with computer-generated works and work-for-hire doctrine, the legal framework for AI-generated inventions could establish clear ownership rights through two potential mechanisms. The concept of “deemed inventorship” would allow human applicants, such as companies or programmers, to be legally recognized as inventors of AI-generated innovations, mirroring how employers own intellectual property created by their employees under traditional work-for-hire principles.
The system alternatively proposes a special procedural exception that permits listing AI as an inventor while ensuring automatic assignment of rights to the human applicant. This addresses the existing legal requirement for inventors to transfer their rights—a capability AI systems inherently lack. Both approaches tackle the fundamental challenge of attributing inventorship in AI-generated innovations while ensuring that commercial and legal rights remain with human entities capable of exercising them.
Because of this shift, the developer of AI would have a greater incentive to create AIs that are more effective and to use those AIs to generate more patentable ideas for the benefit of society as a whole.
There are a number of reasons why it is essential to grant patents to AI-generated inventions. Primarily, grant of patents, acts as an incentive for the inventors to continue inventing new products as they act as a big monetary incentive for them as well as the businesses. If there is no patent, then there will be no monetary incentive for the inventors to develop new products or for the companies to invest in research and development (R&D). Against that, it will give rise to a free-riding problem where the competitors will steal the idea of the innovation and develop the same product at a lower cost as they did not incur any R&D cost to develop their product. Due to this, the original inventor will lose a substantial portion of market share and will put a big dent in the economy.
Secondly, patents can play a crucial role in commercializing AI-generated inventions. With the rise of new technologies, many people are keenly interested in investing in these innovations to further develop them and make gains. Patents give investors an opportunity to invest in such technologies and gain profit as they will get the exclusive right to manufacture, use, and sell their inventions for a limited period. This exclusivity encourages inventors to invest in bringing their inventions to market, leading to greater availability and accessibility of AI technologies.
Further, Patents will pave the way for inventors to disclose their inventions to the public without a free-riding problem by providing a legal framework for licensing and commercialisation (Here). This will also assist them in developing their knowledge by interacting with other inventors, experts, researchers and industry players. This, in turn, can accelerate the development and adoption of AI technologies, benefiting society as a whole.
Current Legal Position in India
Obtaining a patent in India requires meeting several essential criteria outlined in the Indian Patents Act, 1970. Firstly, the invention must be novel, meaning it should not have been disclosed to the public anywhere in the world before the date of filing the patent application. Secondly, the invention must involve an inventive step, meaning it must not be obvious to a person skilled in the relevant field of technology. Thirdly, the invention must be capable of industrial application, meaning it can be made or used in an industry.
In India, Computer-related inventions (CRIs) are examined based on the subject matter exclusions defined in Section 3(k) of the Indian Patents Act, 1970, which proscribes patentability of “mathematical methods, business methods, computer programs per se, and algorithms”. However, AI-based inventions that are novel, non-obvious, and useful can be patentable in India.
Current International Legal Position
At present, most countries, including India, are in a legal quandary regarding patentability of AI-generated inventions. For example, the EPO has clarified that in order to be an inventor, a person must be a ‘natural person’, and therefore an AI system cannot be designated as an inventor. The USPTO took the same stand, where the Supreme Court clarified that an ‘individual’ is understood as a ‘human’ and not a machine. In the US Supreme court judgement in Mohamad v Palestinian Auth., the court ruled that section 115 (b)(2), which is related to inventor’s oath or declaration mentions the term ‘himself’ or ‘herself’ rather than ‘itself,’ it means the law intends for inventors to be real persons. It simply leads to the inference that since AI can’t take oaths, AI can’t be an inventor. The use of ‘whoever’ in U.S. Patent Act (35 U.S.C) doesn’t change this requirement to name a human inventor in a patent application.
The courts don’t appreciate the idea of AI being an inventor. It is more likely that the legislature will have to find a solution by making a new law or changing the patent rules that are already in place. So first, there is a need to shift the basic proposition of law to grant patents to AI-generated inventions. As mentioned in the previous section, granting patents to AI-generated inventions is important to incentivise research and development, monetise new inventions and solve free-riding problems.
However, on the contrary, the Australian patent system initially took a favourable decision for considering the AI machine as an inventor, which was later turned down by their Federal Court. Surprisingly, South African patent office has granted DABUS as an inventor, which is the case gone in favour of considering AI machines as an inventor. Thus, it appears that most of the patent offices do not favour considering AI machines as an inventor. As more and more AI patent applications are filed, there will be more clarity on the issue of inventorship in the coming years.
Therefore, the major question, then, is, who should be granted the patent for AI-generated inventions in India.? The authors have examined different approaches in order to devise a viable approach for Indian Patents system.
Patent to the AI technology?
As per the Patents Act of 1970, only a “person” can get a patent for developing a new product. As AI technologies are not persons, they can’t be granted patents in the current legal framework. If the legislation allows AI software to get a patent, it will create a significant change in the current legislation in order to provide patents to these technologies thereby creating complexity in Indian Patents system. This can also bring it open for misuse by others as the definition of ‘person’ will change the scope and ambit of applicability of the Act and people will be able to get patents without contributing much into an invention.[1]
AI-developer as patent holder?
While AI developers play a crucial role in creating and training AI systems, their direct involvement in specific inventions generated by these systems can be minimal. The “black box” nature of AI systems makes it difficult to determine whether the developer or the operator had more substantial involvement in the final invention. This ambiguity can disincentivize human inventors who rely on AI for assistance, as they might not receive due recognition for their contributions. The challenge lies in ensuring that the patent system fairly acknowledges the human element in AI-generated inventions, whether it stems from the AI’s developer or its operator. Ensuring a balanced approach that recognizes the roles of both parties is essential to foster innovation without granting undue advantages.
AI operator as patent holder?
The individual who operates the AI and provides the input or “prompt” shall be considered the inventor by the authors for the limited purposes of this section (It is based on the assumption that the inventor will be the person who will operate the AI or he will hire someone to operate AI, in which case the inventor will remain as the ultimate owner of the AI-generated invention). This aligns with the principle in Indian patent law that the inventor is the person who conceives the invention. While the AI may autonomously generate the invention based on the input, the operator’s role in directing and controlling the AI’s actions is crucial. Granting the patent to the operator incentivises them to use AI technology innovatively while ensuring that human creativity and decision-making are recognised and rewarded.
Significant oversight approach
As discussed above, there is a need to have a shift in the legislation in order to grant patents to AI technology. However, making a general rule to provide patents to the AI operator may turn out to be a tragedy as it will lead to expansion in the definition of ‘person’ in the act. Therefore, the authors propose a new approach whereby human contribution in an invention must become one of the significant principles, alongside novelty, non-obviousness, and commercialization. Patents for AI-developed products should be granted only when there is a significant contribution by the inventor in developing the new product.
The “innovation oversight” approach proposes a framework for providing patents to the inventors only when there is a significant intervention from inventors. This approach seeks to address the challenges posed by the autonomous nature of AI technologies and ensure that patents are granted to individuals or entities that exercise intellectual control over the invention. As there is no precedent in India with regard to the patentability of AI-developed products, reliance can be placed on USPTO Inventorship Guidance and Thaler v. Vidal to form the basic premise of the framework which provides that AI-generated inventions can be patented only when there is a significant human contribution. The authors have blended this premise with Indian Patent framework to derive the “Significant oversight approach”.
The “Innovation Oversight” approach establishes a structured three-step test to determine patent eligibility for AI-generated inventions: first, identifying the extent of human intervention in the AI’s operation; second, evaluating the intellectual contribution and value addition by the human inventor; and finally, establishing legal ownership based on these contributions. This systematic framework ensures that patents are granted only when there is substantial human oversight and intellectual input, while also providing clear guidelines for resolving ownership disputes between inventors and operators.
1. Identifying Human Intervention
The first stage of applying this approach is to identify the human intervention of the human operator in order to develop and control the AI technology. The inventor shall be responsible for making decisions that influence the outcome of AI-generated inventions. The inventor should provide inputs, set parameters, and make adjustments to the AI. There might be cases where the inventor who possesses the non-obvious skills and the operator who is operating the AI technology are separate, leading to disputes regarding patent ownership. Ideally, the inventor should receive the patent in such disputes as the new AI-generated invention will be formed due to non-obviousness in the invention which can only be brought by an inventor and not operator, but these cases must be decided on a case-by-case basis, considering the importance and contribution of the prompt in the development of AI products.
2. Determining Intellectual Contribution
Once the role of the human operator is determined, the next stage will be to determine the intellectual control of the inventor over the new product. The most important part of determining a product’s patentability is the inventor’s value addition. So, it is of utmost importance for the courts to determine whether the basic value addition came from the intellect of the inventor or AI technology. This involves assessing whether the operator’s decisions and inputs have resulted in a novel and non-obvious invention.
The inventor’s role in identifying and providing prompts, as well as making adjustments to the AI system to achieve the desired outcome, are key considerations in this evaluation. Based on the significant intellectual contribution, it must determine whether another innovator with the same level of skill could have developed the same product from that AI. If there is a significant intellectual contribution from the inventor, then the inventor must be entitled to get a patent for the new product.
3. Establishing the Legal Ownership
Based on the evaluation of the human operator’s intellectual control and inventive contribution, the legal ownership of the patent is established. If the operator has exercised sufficient intellectual control and made a significant inventive contribution to the AI-generated invention, they are deemed the rightful owner of the patent. This approach ensures that patents are granted to individuals or entities that actively contribute to the inventive process ra ther than to the AI technology itself.
In the case of different inventors and operators, if the contribution of the prompt was significant and the inventor could not have produced the product without the operator, then the ownership must be shared in proportion to their contributions under Section 50 of the Patent Act. In such cases, the proportionality test must be applied, taking into account the contributions of both the innovator and operator, and the ownership must be shared accordingly. Even in cases where the patent is licensed out for commercial use, the revenue from such license shall be distributed again in proportion to the contribution of the inventor and operator.
Conclusion
The rapid advancement of artificial intelligence presents both opportunities and challenges for intellectual property rights, particularly in the realm of patent law. The current lack of clarity in Indian patent law regarding the patentability of AI-generated inventions underscores the need for reform to ensure that innovation in AI is appropriately incentivised and protected.
The “Innovation Oversight” approach proposed in this article offers a pragmatic framework for evaluating the patentability of AI-generated inventions. By focusing on the overall inventive concept and the contribution of human ingenuity, this approach seeks to strike a balance between rewarding innovation and fostering access to AI technologies. Moving forward, it is essential for policymakers, legal experts, and industry stakeholders in India to engage in dialogue and collaboration to develop a robust framework that addresses the unique challenges posed by AI. By doing so, India can position itself as a leader in AI innovation while ensuring that the benefits of AI are realised by all.
[1] Adam Lidgett, ‘USPTO Tells Fed. Circ. Only Humans Can Be Inventors, LAW360’ (Feb. 4, 2022) <https://www.law360.com/articles/1461934/uspto-tells-fed-circ-only-humans-can-be-inventors> accessed on 10 May, 2024
*Harshal is a third-year student pursuing his B.A., LL.B. (Hons.) from Gujarat National Law University.
**Arihant Sethia is a third-year student pursuing his B.Com LL.B. (Hons.) from Gujarat National Law University.
Comments