CAGE


Written by Rohaan Thyagaraju V, a fifth-year B.B.A. LL.B. student at Symbiosis Law School, Hyderabad.

  1. Introduction

India’s lightning-fast progress towards artificial intelligence shows the increased importance of this technology from an economic and strategic perspective. According to NASSCOM, AI-related processes and intelligent service solutions can add around USD 500 billion to India’s economy by 2025, making the country one of the leading players in the global AI economy. At the same time, AI has changed the whole way digital content has been created, and the interface has shifted from traditional software to natural languages. According to expert projections, in 2024, the global generative AI market will be valued at USD 44 billion, which is expected to rise to over USD 200 billion by the year 2030. At the core of this transformation lies the technology of prompt engineering, which creates specific instructions permitting large language models (“LLMs”) like OpenAI ChatGPT, Google Gemini, and Anthropic Claude to produce the best results. Prompt engineering is already in demand in many fields, including technology, marketing, law, education, and software engineering. This commercial shift raises a significant legal question under Indian law: Can prompt engineering be protected as intellectual property?

Very few would dispute that the issue at hand is technological, intertwining aspects of ownership, innovation incentives, job mobility, platform governance, and market concentration. India’s Intellectual Property (IP) laws largely adhere to traditional categories such as literary works and patents, but they do not adequately encompass the emerging field of prompt engineering. The classification of prompts varies depending on their creation, application, operation, and commercialisation, leading to partial protection under existing laws such as copyright and trade secrets Act. However, there is no comprehensive legal framework for the diverse functional and expressive qualities of prompts. 

Prompt engineering is more than an ordinary text

In most cases, using an AI-based prompt to complete a task is as simple as typing a command into an AI program. However, prompt engineering with advanced features often requires many iterations and tests to improve the machine’s performance.For example, a legal technology start-up in Bengaluru might create proprietary prompt sets to generate standardised drafts of shareholder agreements that comply with both the Companies Act, 2013, and SEBI disclosure requirements. Alternatively, an e-commerce company in India might develop prompts specific to its recommendation engine and track consumer usage patterns for analytical purposes.

  1. Application of the Copyright Act, 1957 for prompt engineering

The applicability of the Copyright Act of 1957 to prompt engineering deserves closer attention.  As per Section 13, which defines literary works and establishes their originality. This also applies to prodigal prompts as an instance of text-based literature. 

The definition of original literature has also been expanded through numerous judicial interpretations across various levels of Indian courts, further reinforcing the limits of defining what constitutes original literature. For example, the Supreme Court in “Eastern Book Company v. D. B. Modak (2008)” stated that originality is determined by answering the question of whether a work incorporates a creative concept and the degree of one’s own individual effort, rather than being determined solely by the amount spent in producing the work. 

This decision has fundamentally altered how copyright law is applied in India since that ruling, making it reasonable to conclude that a prompt designed using higher forms of linguistic artifice could meet the threshold for originality as defined in the Copyright Act. Therefore, one could create a prompt design that utilizes a higher degree of complexity and sophistication than a generic prompt design.

  1. The Idea-Expression Dichotomy

Copyright law in India protects the expression of an idea but not the idea itself. In R.G. Anand v. Deluxe Films [1978], the Supreme Court of India found that copyright does not protect a theme/concept. This presents a challenge for prompt engineering. A prompt is a command that directs a user to produce something. An example of a prompt is “Please summarise this case in layman’s terms.” Copyright does not protect this prompt, as it is simply a command, unlike a work of original expression. 

The Copyright Act, 1957, provides copyright protection for software in India under Section 2(ffc); however, it applies only to expressions of software in source code, not to abstract or functional software. As with software copyright, only the express component of a prompt may be protected by copyright; its functional aspects are not protected.

  1. Merger doctrine and functional constraints

The merger doctrine poses yet another barrier. The limited number of ways to express functional instructions implies a thin chance that an expression will come into agreement with the underlying idea. Most prompt designs have a standardised structural configuration. 

For example, when coded consistently, legal analysis or image-generation scenarios generated by this technology will likely exhibit a predictable pattern. Overprotection can monopolise words in common speech, making it difficult for others to generate functional language. Therefore, if someone attempted to claim an exclusive right to use prompts to generate employment contracts under Indian labour law, the rest would be legally limited in their interactions with the technology’s functional aspects.

  1. Prompt libraries and database protection

Curated prompt libraries could be afforded higher levels of copyright protection than their uncurated counterparts if they are considered compilations as per Section 2(o) of the Copyright Act,1957. The rationale is that they contain an element of creativity in the choice of works. For example, a company may have innumerable organised prompts by industry and use a type which is coupled with an optimised output. However, competitors will still be able to create similar prompt systems independently, as long as those systems are developed differently.

  1. Patent Law and AI Instruction Systems

Patent laws can be appealing due to the quantifiable technology enabled by rapid systems, but Indian patent laws are very restrictive. Section 3(k) of the Patents Act, 1970 excludes computer programmes, algorithms, mathematical methods, and business methods from patentability. As such, there would be no possibility of a prompt being granted patent eligibility due to the algorithmic interaction with AI systems, as the means by which most prompts produce an optimised computational result without changing the computer hardware architecture. Therefore, a prompt would not meet the requirements for patent eligibility. 

III. Alternate IP Protection for prompt protection

The Delhi High Court decision in Ferid Allani v. Union of India (2019) ruled that computer-related inventions which are deemed to provide a technical contribution should not be automatically excluded from patenting simply because they use software. This decision provides a limited opportunity for patent claims concerning advanced prompt architectures that are part of or fully integrated into a technical system. To quote an example, a medical AI prompt system improving diagnostic accuracy or a cybersecurity prompt framework detecting phishing patterns.

Similarly, in the past, Indian courts have always found that protectable confidential commercial information exists. The Delhi High Court previously protected technical know-how from being disclosed without prior authorisation in the case of John Richard Brady v. Chemical Process Equipments Pvt. Ltd. (1987). 

  1. Public interest limits of prompt engineering

As companies seek quick protection, it may create significant problems for agencies, potentially leading to expanded proprietary rights. The objective of intellectual property law is not only to provide incentives for innovation but also to reward individuals’ creativity and establish an equilibrium between stimulating innovation, allowing for public access, and enabling healthy competition within the marketplace.

  1. Risk of market concentration in emerging AI economy 

The AI economy in India largely relies on startups, open-source developers, and education and innovation performed at low or no cost. As stated in the NASSCOM 2023 report, India has over 4,500 AI startups that generally have very limited access to funding when compared to multinational corporations. Overprotecting prompts will create a large windfall for the dominant technology players, who can build their own proprietary ecosystems to handle prompts, and will also create additional barriers for smaller innovators to access. This goes against India’s broader digital development goals.

  1. The practical limits of enforcing prompt rights 

Even if courts protect legal rights to some extent, it will be very challenging to enforce. Minor changes to wording would often yield similar outputs, thereby inundating the court with challenges of evidentiary questions such as: How similar do the prompts need to be for infringement? Is similarity determined by some form of functionality or by wording? Do rights of ownership arise when the AI-generated output from the prompt is optimised into a better form? How do the courts determine the open-ended cycles of prompt evolution? India has faced extensive institutional challenges in software and copyright litigation, and prompt disputes may create significant technical complexities.

  1. India’s Regulatory Window

To date, no jurisdiction has developed a comprehensive regulatory framework for prompt engineering. The US Copyright Office has repeatedly stated that copyright protects human-authored content but has done little to clarify how copyright would apply to AI-generated content. Thus, prompt-specific or prompt-related jurisprudence still needs to be clearly defined.

The European Union’s AI Act primarily focuses on oversight, risk categorisation, and accountability rather than prompt ownership. Because of this, India has an opportunity to establish a balanced regulatory regime before monopolistic tendencies become institutionalised. Indian intellectual property policy has historically maintained a developmental balance approach.

This approach is visible in aspects such as compulsory licensing under pharmaceutical patent law, restrictions on software patents, fair dealing protections in copyright law, and judicial emphasis upon public interest. Thereby, prompt engineering should be approached with similar caution.

IV. Toward a Balanced Indian Framework

Revising the intellectual property law is not necessary for prompts in India; rather, an effective response would be to take a calibrated approach as follows: 

Copyright protection should be limited to compilations or original prompts

India should establish a clear legal framework to protect trade secrets, ensuring enterprises have legal protections for their confidential assets and systems. AI platforms must also be required to disclose the following:

  1. whether prompts are being stored,
  2. whether they will be used for training,
  3. whether the platform was confidential,

If significant market share owners create proprietary prompt ecosystems creating barriers to competition, the Competition Commission of India (CCI) may need to collect data and take additional regulatory action under the Competition Act. The CCI may also need to deal with issues of AI platform dominance and data concentration. Finally, institutional capacity building will also be needed to support and create the institutions necessary for addressing prompt engineering disputes involving computing, linguistics, and Software.

Conclusion

Prompt engineering offers a clear example of the difficulty that generative artificial intelligence poses for existing legal classifications in India. Current Indian laws do not explicitly cover prompts, as their use does not clearly fall under copyright, patent, or trade secret law. 

Protection of prompt engineering through trade secrets is likely the most viable option. However, trade secret enforcement across jurisdictions varies widely. The primary issue is whether intellectual property law governs these interactions between man and machine. However, negligible protection may decelerate advancement, owing to reduced investment in developing new products and services. The Indian government, therefore, needs to implement a balanced approach, enhancing the country’s current technological capabilities without giving up the right to sole proprietary ownership of a technology or product. The discussions regarding prompt engineering will not be restricted to the owner of the prompt but also include how companies would establish future associations with artificial intelligence. 

 

Leave a Reply