Introduction
As Artificial Intelligence (AI) quickly spreads throughout every aspect of human life, there are many legal challenges that can’t currently be solved because no law has been made for them. One of the most important issues is whether or not prompts (the instructions we give to AI systems) can be protected legally under copyright law as intellectual property rights (IPR). This isn’t just a theoretical problem; there is so much money being made by companies working with AI technologies that who owns the outputs produced from those technologies (and the inputs used to create those outputs) has become commercially important. Today in India, which has been rapidly establishing itself as a top AI nation in the world, there appears to be no laws about this subject.
Prompt engineering is the process of developing detailed and specific prompt-based instructions to direct AI systems in order to produce certain results. Prompt engineers who are highly skilled take a lot of time and thought in creating prompts to generate advanced responses from large language models (LLM) such as GPT-4, Claude, and Gemini. The prompt might include a number of examples, several types of directional prompts arranged in a logical sequence, a definition of the role that the LLM will play, constraints that need to be followed by the AI, and a sequence of logical steps. All of these items created together constitute a creative and technical work that clearly has the essence of an author. [1]
The Copyright Act of India of 1957[2] (hereafter referred to as “the Act”) is the primary copyright legislation for copyright protection in India. The Act went into effect long before generative AI was developed, but the language of the Act is broad enough and the courts have interpreted it in a flexible enough manner to cover future forms of intellectual creation. The question of this research is whether AI-generated prompts that have a distinct design and an original nature are eligible to be called an “original literary work” under the Copyright Act, thereby making the AI-generated prompts qualify for copyright protection.
This article is divided into six parts. The first part is an introduction; the second part highlights the concept of originality as defined by Indian Copyright Law and discusses leading case laws. The third part looks at how to apply the idea/expression dichotomy to AI prompts and provides taxonomy of prompts along a copyrightability continuum. The fourth part addresses the issue of authorship and how AI generated output challenges traditional definitions of human creativity. The fifth part of the article provides policy recommendations for the creation of a comprehensive legal framework for prompt engineering in India, and the final part is the conclusion.
Originality under indian copyright law: the “modicum of creativity” standard
The foundation of copyright protection in any country is the requirement of originality for a work to be eligible for copyright protection. In India, the Copyright Act does not explicitly provide a definition of originality; however, it states that for a work to be eligible for protection under section 13[3], it must be ‘original’. Over time, Indian judges have developed a sophisticated understanding of this concept, which culminated in the Supreme Court’s ruling in the Eastern Book Co. v. D.B. Modak [4] case; this decision has become the leading authority on originality in India.
Before the Eastern Book Company case (2008), Indian courts had adhered to the “sweat of the brow” principle taken from English law, which stated that anything made from hard work and labor would get copyright protection, regardless of how creative it was. The Privy Council’s decision in the Macmillan & Co. Ltd. v. K. & J. Cooper[5] case, was cited many times as stating that a person’s own labor invested in selecting and arranging material was sufficient to give it creative originality, although this interpretation was considered by many to go well beyond the primary intent for providing copyright protection, which was to encourage actual creative contributions.
The Eastern Book Co. case saw the Supreme Court of India set aside the “sweat of the brow” standard and adopt an approach to originality that is much closer to that used by the U.S. Supreme Court in the Feist Publications, Inc. v. Rural Telephone Service Co.[6] case. The Court found that the level of creativity required is not very high and that the author must have added some form of creative talent or ideas to the work to avoid it being a mechanical reproduction or simply having been done with effort. The Copyright Office has developed a middle road between the rigorous “genius” test of some continental legal systems and the sweat of the author’s brow doctrine, which is the reasonable modicum of creativity standard. As the Court stated, that an author seeking to obtain a valid copyright registration in a compilation of works must create the compilation in a way that involves creativity. The creation consists of more than merely making trivial variations or creating only minimal amounts of creative work; each author must contribute a minimum of creativity in order to meet this standard.[7]
When applied to AI prompts, the reasonable modicum of creativity standard provides insights. A prompt engineer, who carefully constructs a hierarchy of nested instructions, defines persona types, defines logical environment constraints, and uses an appropriate contextual scaffold to provide prompt context/meaning is not merely mechanically reproducing data or information. The engineer who produces an AI prompt is deliberately selecting the structure, order, emphasis, and manner of conveying the information contained in a prompt; those are choices involving judgment, skill, and a reasonable amount of creativity. Importantly, prompt selection, prompt arrangement, the creation of prompt-based examples (few-shot prompting), and the calibration of tone and scope each involve a degree of intellectual discretion that the courts have generally considered creativity under the law.
The Act states that a “computer programme, tables and compilations including computer databases” are types of literary works covered by the Act.[8] Prompts that direct how an AI will perform operations are similar to computer programmes, as both have the same purpose of directing machine behavior through text-based instructions using precise logical sequencing. The structure of both requires a significant investment of time and creativity. While both types of directive documents are similar in terms of originality, the manner in which they are written and the language used to do so clearly separates them from each other.
In Navigant Consulting (PI) LLC v. Wipro Ltd.[9], the Delhi High Court held that originality in professional and technical writing requires the use of skill and judgment, not just a result of hard work. Additionally, in Anil Gupta v. Kunal Dasgupta[10], the Court said that even an outline (also known as a template) or a concept for a television programme could be protected if it demonstrated original authorship. These cases suggest that even if creative works are expressed using unconventional or new forms of media, there is still a possibility of obtaining copyright protection as long as the work meets the level of required originality or authorship.
Idea-expression dichotomy & taxonomy of prompts
A. The Idea-Expression Dichotomy in Indian Law
Copyright law is built on the concept of distinguishing between an idea and how that idea is expressed. Copyright will protect the way an idea is expressed, rather than the actual idea itself. This distinction is inherent in the Anglo-American copyright tradition and consistently recognized by Indian Courts as well, and this principle has significant consequences for the copyrights of AI generated prompts. This distinction was confirmed by the Indian Supreme Court in the case of R.G. Anand v. Delux Films[11], where it was stated that copyright only existed in the medium through which a work was created, and not in the underlying theme, plot, or subject matter.
The beginning of the distinction between an idea and its expression in common law comes from the US Supreme Court case Baker v. Selden[12]. Since then the standard has been universally adopted in copyright law. This standard was first developed in the UK courts, and was addressed in cases such as Kenrick & Co v Lawrence & Sons[13] and later enforced in the Court of Appeals in the case of Hawkes & Son (London) Ltd v Paramount Film Services Ltd[14]. Again, the concept of an idea vs. an expression has been consistently applied in the Indian courts, which also recognize that the distinction between idea and expression is not always easy to ascertain and must be reviewed on the facts of the case.
The application of the idea-expression dichotomy to AI prompts requires careful thought and analysis. The most basic way to define the concept behind an AI prompt is to refer to it as an idea (directing an AI to create a specific output type). At this level of abstraction (the most general), it is obvious that a prompt’s idea cannot be protected. However, as the prompt moves toward being defined (specific) through the use of language, structure, sequence and/or logic, this prompt transition from an idea-only prompt to an expression of that idea prompt. Therefore, the crucial question is what point along this continuum a prompt crosses from an unprotectable idea to a protectable expression.
B. Taxonomy of Prompts Along the Copyrightability Spectrum
The following taxonomy of AI prompts based on their copyrightability has been established based on the previous analysis:
- Minimal Prompts: Prompts consist of one or a few words or very short phrases such as “write a poem on love,” or “summarize this article.” These prompts do not have any individual expression. These prompts represent an idea at its most abstract level and do not contain enough fundamentally unique creative choice (the “modicum of creativity”) to be copyrightable. They are similar to titles, short phrases and other works that almost always are not copyrightable.
- Intermediate Prompts: These prompts also include a minimal number of guidelines but have more detail to them. For example, the following would qualify as an Intermediate Prompt: “Write a 500-word blog post in a conversational tone aimed at millennials in India on the benefits of meditation and use a strong headline followed by 3 subheadings”. This prompt contains an element of individual creativity and should fall under the modicum of creativity standard. The copyrightability of intermediate prompts will need to be evaluated on a case-by-case basis.
- Complex and Structured Prompts: This type of prompt is most likely to get copyright protection. A complex prompt will include factors that make up the persona for which you are creating detailed descriptions. Further, it will include several examples of use, known as few-shot prompts and have explicit instructions on chain of thought (i.e., reasoning). You will provide the AI with a framework for establishing its reasoning and give it very specific constraints for tone, format, and content. The amount of intellectual investment in creating such a prompt is considerable, and the author’s creative judgment is evident based on the decisions made in developing their prompt. The author believes that such prompts will qualify as original literary works for copyright protection.
- System Prompts and Proprietary Prompt Libraries: Many organizations develop complex systems that use a specific system prompt to define the behavior of an AI assistant within an entire product. System prompts can be hundreds or thousands of tokens long, and they demonstrate substantial creativity and technical knowledge in their development. In effect, they provide the “personality” and “intelligence architecture” of the AI system when it is deployed. As such, these prompts clearly exhibit substantial intellectual investment and creative expression, and therefore, they should easily qualify for copyright protection.
The Act also affords protection against the use of “compilations” pursuant to Section 2(o) [15] when a prompt consists of a curated collection of examples, instructions, and contextual materials that have been integrated to form a single unit. A compiler of a database is entitled to protection based on the original selection and arrangement of the contents of the database, and similarly, therefore, a prompt engineer who makes original selections with respect to assembling the constituent parts of a complex prompt will be entitled to obtain protection based on that selection and arrangement of the component parts of the prompt, even though none of the individual constituent parts will be entitled to independent protection under the Act.
Authorship, human creativity, and the challenge of ai-generated outputs
The determination of authorship is an essential component of an analysis of copyright. In the U.S. Copyright Act, copyright is granted to an “author” of an original creation. The Act also provides a specific definition of authorship in the context of computer-generated works, stating that the “author” of a computer-generated work is “the person who causes the work to be created.” This definition is significant to the issue of AI-prompts in two respects. First, it defines the human author of the prompt as a potential author of the AI-generated creation; second, it raises the question of the identity of the human author of the AI-prompt.
With respect to the first issue, the definition of “author” for computer-generated works in the Act is a statutory basis for granting copyright to the prompt engineer or user who caused the AI to generate a specific result. This interpretation is complicated by the fact that the AI output is not generated directly by the prompt but is instead produced as a result of an intermediate process between the human prompt and the AI output. The AI operates through a system of billions of parameters, which are created through machine learning based on the training data applied to the AI system. The extent of human control over the output of AI is subject to much variation depending on the complexity of the task and the AI system’s overall architecture.
Several courts have considered these matters in other jurisdictions. In Thaler v. Vidal[16], the United States Federal Circuit refused to allow an inventorship claim based on an artificial intelligence (AI) inventor and instead required that the inventor be a person. Similarly, the U.S. Copyright Office has determined that it will not provide registration for works created by a machine or entirely mechanical process without a human author. [17] In Naruto v. Slater[18], the Ninth Circuit denied copyright protection to a non-human primate on the grounds that only human authors are entitled to protection under the Copyright Act. While Indian courts have not yet directly addressed the issue of AI as an author under intellectual property laws, these decisions should have persuasive value.
The second question about copyright with a focus on the “prompt” and whether it can be copyrighted and how the work the copyright pertains to (the “output”) can be viewed as an easy answer. The prompt is clearly a human authored text because it was written by someone using their own judgment and creativity to form words into a sentence, so it is a written work that has no AI output. AI output has nothing to do with how the creator of the prompt created the prompt. As long as the prompt meets the originality requirements described in Part II, the creator of a prompt is unequivocally the author of that written work and has the right to copyright it.
Thus, there is a clear difference from a legal point of view and from a practical point of view on the copyright of the prompt (the human written input into AI) versus AI generated output (machine made content). Therefore, the prompt may not have an established copyright, but will have an established copyright for a prompt engineer as an independent literary work. Therefore, the prompt will stay the intellectual property of the individual that created it, regardless of the AI uses it to create additional works.
An additional question regarding authorship relates to who owns and can license rights in prompts. According to the Act, in circumstances where an employee creates a prompt within the scope of their employment, copyright in that prompt (i.e., the creator of the work) will automatically vest in the employer unless there is an agreement to the contrary. Likewise, copyright ownership for a prompt created by an independent contractor will depend upon the applicable contractual relationship between the prompt creator and the entity that retains them through that contract. As illustrated, this is not simply a theoretical issue; to give just one example, most commercial AI development companies employ multiple prompt engineers, generally under employment contracts. Therefore, establishing ownership over the prompt libraries that are generated from this work could have significant commercial implications.
The case for legislative reform: toward a framework for prompt engineering in india
The foregoing analysis demonstrates that the existing provisions of the Indian Copyright Act, 1957 are capable of being interpreted so as to extend protection to sufficiently original AI prompts. Nevertheless, the absence of explicit statutory recognition creates legal uncertainty that is inimical to innovation and investment. India’s National Intellectual Property Rights Policy, 2016[19] expressly acknowledges the need for an IP ecosystem that promotes and protects innovation. A coherent legal framework for AI prompts would be an important step toward realizing this objective.
The current legal lacunae are several. First, the Act does not explicitly address AI prompts or the outputs of generative AI systems, leaving courts to reason by analogy from provisions designed for very different contexts. Second, the authorship provisions, while arguably extensible to AI-generated works, are ambiguous in their application and have not been tested in Indian courts. Third, there are no guidelines or safe harbors for the use, reproduction, or commercialization of AI prompts, creating uncertainty for both prompt engineers and the businesses that deploy their work. Fourth, the increasing tendency of AI companies to claim ownership of all prompts submitted to their platforms through terms of service agreements[20] raises questions about the enforceability of such claims under Indian contract and copyright law.
Using the European Parliament’s activity regarding AI liability[21] and the work of international scholars studying AI and copyright[22], an additional set of recommendations for drafting a legislative framework is as follows:
- The Act should include an explicit definition of “prompt” and clarify that prompts constitute “literary works” pursuant to the Copyright Law where prompts contain enough originality. Doing this will create certainty in the minds of prompt designers and remove any ambiguity regarding whether existing laws are relevant.
- The Act should also have provisions preventing copyright protection for lower tier prompts (i.e., simple prompts) and provide complete copyright protection for higher tier prompts (i.e., complex and structured prompts). This is consistent with currently existing laws governing copyright protection provided for compilations and databases as defined by Section 2(o); and it aligns with the creativity standard as established by the United States Supreme Court.
- The authorship provisions of the Act should be clarified as to their specific application to works produced by artificial intelligence and their relevant human prompt engineers who provided direction for the AI program’s output. A well-defined statutory authority would set the expected parameters of when a human prompt engineer will qualify as the creator or author of output that was generated from artificial intelligence. This authority would also create guidelines for the sharing of copyright among all contributing authors or parties who created both the AI program and the outputs produced by the AI program.
- The Government should work with members of each sector to draft additional guidelines for prompt engineering that are relevant to AI programs and their outputs comparable to the Software Industry Guidelines that are produced by NASSCOMM.[23] These industry-specific guidelines for prompt engineering would cover matters of originality, authorship, as well as the assignment and licensing of related copyrighted works produced by AI programs. Those guidelines could also be created in collaboration with others in the technology industry including stakeholders, experts from academia, and lawyers in the area of technology law. Such guidelines would also be expressly consistent with the objectives expressed in India’s National Strategy for AI[24] which sees India becoming the world’s leader in responsible development of AI technologies.
- The Government should reach out to the World Intellectual Property Organization (WIPO) and other international organizations to promote the development and implementation of global standards for the protection of intellectual property rights related to artificial intelligence. Indian scholars have pointed out that both the regulation of AI and copyright are international issues that will require coordinated international action to resolve. Because India is one of the largest developers of artificial intelligence technologies globally, it has both a vested interest and the ability to participate in this effort.
Conclusion
The rise of prompt engineering as a professional field and an emerging economic driver requires a consideration of a number of questions that require resolution within the context of Indian Copyright Law. The Indian Copyright Act of 1957 provides a legal basis for protecting the sufficiently original and creative AI prompts under copyright law, on the grounds of originality, literary works, and authorship. A sound basis for such analysis can be found in the “modicum of creativity” standard established by the Supreme Court in the case of Eastern Book Co v. D B Modak, which establishes the appropriate balance between requiring a threshold of creativity while also ensuring that valid, creative contributions are not excluded by way of imposing an overly demanding threshold upon applicants for copyright protection.
The idea-expression dichotomy within the foundation of Indian Copyright Law can also be utilized as a principled means for differentiating between non-protectable simple/ generic prompts and protectable complex prompts. Indeed, there will be many simple/generic prompts that convey an idea in its most abstract form which will not qualify for copyright protection. Conversely, however, carefully constructed prompts that have been created using original arrangement, selection, and expression of linguistic/structural elements, will, in my opinion, fall within the domain of copyright protection as original literary works.
The question of authorship is more complicated when it comes to AI-generated outputs, but relatively straightforward when it comes to the prompt itself: the human creator of a prompt is its author and is entitled to assert copyright in it as a distinct literary work. This principle, rooted in the Act’s own provisions on computer-generated works, allows prompt engineers to protect their intellectual investment while the wider question of copyright in AI outputs remains uncertain.
But in the end, legislation will be needed to create effective and clear legal protection for AI prompts. The existing statutory regime can be extended through interpretation but was not enacted for the digital era of generative AI. India has an opportunity to take the lead globally by making a forward-looking amendment to the Copyright Act that deals explicitly with prompt engineering, establishes a leveled standard of copyrightability, clarifies the rules of authorship, and lays the groundwork for sector-specific guidelines. Such reform could improve India’s broader strategic interests in developing a thriving and innovative AI ecosystem as well as protecting the legitimate interests of prompt engineers and the companies they work for. The prompt is property and Indian law must say so.
*****
Footnotes
[1] World Intellectual Property Organization (WIPO), WIPO Technology Trends 2019: Artificial Intelligence (2019), available at https://www.wipo.int/edocs/pubdocs/en/wipo_pub_1055.pdf
[2] The Copyright Act, 1957, No. 14, Acts of Parliament, 1957 (India).
[3] The Copyright Act, 1957, Sec 13(1)(a) (India).
[4] Eastern Book Co. v. D.B. Modak, (2008) 1 SCC 1(India).
[5] Macmillan & Co. Ltd. v. K. & J. Cooper, (1924) 26 BOMLR 292 (India).
[6] Feist Publications, Inc. v. Rural Telephone Service Co., 499 U.S. 340 (1991).
[7] Eastern Book Co. v. D.B. Modak, (2008) 1 SCC 1, 93 (India).
[8] The Copyright Act, 1957, Sec 2(o) (India) (defining ‘literary work’ to include computer programmes, tables and compilations including computer databases).
[9] Navigant Consulting (PI) LLC v. Wipro Ltd., 2020 SCC OnLine Del 1240 (India).
[10] Anil Gupta v. Kunal Dasgupta, 2002 (25) PTC 1 (Del) (India).
[11] R.G. Anand v. Delux Films, AIR 1978 SC 1613 (India).
[12] Baker v. Selden, 101 U.S. 99 (1879) (establishing the idea-expression dichotomy in U.S. copyright law).
[13] Romer LJ in Kenrick & Co v Lawrence & Sons (1890) 25 QBD 99 (holding that copyright in a drawing protects its particular expression, not the underlying idea).
[14] Hawkes & Son (London) Ltd v Paramount Film Services Ltd, [1934] Ch 593.
[15] The Copyright Act, 1957, Sec 2(o) (India).
[16] Thaler v. Vidal, 43 F.4th 1207 (Fed. Cir. 2022).
[17] Compendium of U.S. Copyright Office Practices § 313.2 (3d ed. 2021) (clarifying that the U.S. Copyright Office will not register works produced by machines or purely mechanical processes), available at, https://www.copyright.gov/comp3/docs/compendium.pdf
[18] Naruto v. Slater, 888 F.3d 418 (9th Cir. 2018).
[19] National Intellectual Property Rights Policy, 2016 (India), available at https://www.dpiit.gov.in/static/uploads/2025/07/442648f225695b31b259e9d9edd0a8b1.pdf
[20] OpenAI, Usage Policies (2024), available at, https://openai.com/policies/usage-policies (last visited Apr. 1, 2025).
[21] European Parliament Resolution of 20 October 2020 with Recommendations to the Commission on a Civil Liability Regime for Artificial Intelligence, 2020/2014 (INL), available at, https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=oj:JOC_2021_404_R_0006
[22] Andres Guadamuz, Artificial Intelligence and Copyright, WIPO Magazine, Oct. 2017, at 14. Available at, https://www.wipo.int/en/web/wipo-magazine/articles/artificial-intelligence-and-copyright-40141
[23] NASSCOM, The developers playbook for responsible AI in India (2024), available at https://nasscom.in/ai/pdf/the-developer’s-playbook-for-responsible-ai-in-india.pdf
[24] Ministry of Electronics and Information Technology, National Strategy for Artificial Intelligence (2018) (India), available at https://www.niti.gov.in/sites/default/files/2023-03/National-Strategy-for-Artificial-Intelligence.pdf