The Copyright Mirage - Chasing Ownership In Machine-Made Masterpieces
Author: LawSphereX
Category: Intellectual Property
Published:
Updated:
THE COPYRIGHT MIRAGE: CHASING OWNERSHIP IN
MACHINE-MADE MASTERPIECES Utsav Anand, Chanakya National Law University Shailya Sharma, Chanakya National Law University Who owns an AI-generated masterpiece the developer, the prompter, or no one? India's Copyright Act 1957 offers no clear answer. From the RAGHAV episode to the landmark ANI v OpenAI case, this blog unpacks the legal vacuum and the urgent reforms needed to protect creators. Keywords: Generative AI, Copyright Authorship, Indian Copyright Act 1957, AI-Generated Content, Intellectual Property Rights (IPR) Introduction The battle over copyright for AI-generated masterpieces is an emerging issue that requires a creative, urgent, and impactful solution.
Copyright under Intellectual Property Rights is a mechanism that protects the creativity of an individual, whether it is a masterpiece with brushes, a musical rhythm, a song, a mathematical problem, or any other innovative piece of art that holds significance for its author. 1 But the problem arises when your art is being used and modified, and the copyright is acquired by a person who is not a person in reality, i. e. , Generative AI tools, and the contention is that the AI also does not hold the copyright of the modified artwork. Then the question arises, who holds the copyright of those masterpieces? There are a lot of assumptions that are emerging about the ownership of the copyright of AIgenerated content. Unfortunately, in India, the IPR laws do not mention who will be the owner of the AI-generated content.
It is important to analyze and provide a solution, or at least bring this issue to the notice of the government. With the continuing development of generative AI, the copyright claims are getting more complex, which was already a complex phenomenon under the Copyright Law. AI tools like Midjourney, DALLE 3, Google Imagen, and Chat GPT are continuously developing their engine, which are increasing their databases, and the modification of content is at higher risk, so is its originality and copyright.
1 The Copyright Act, 1957, § 13(a) (India).
Generative AI: A Tool that reimagines Generative AI is the tool that generates images, music, poetry, animations, and masterpieces of different art forms based on various inputs. The Generative AI tool uses neural networks to identify the patterns and structures within existing data to generate new and original content. These neural networks are of various types, such as Variational autoencoders (VAEs)2, which, when given an input, an encoder converts it into a smaller, denser representation of the data. Generative Adversarial Networks (GANs)3 were considered to be the most commonly used methodology before the recent success of the diffusion models and Transformative Networks, which are similar to recurrent neural networks; transformers are designed to process sequential and non-sequential input data non-sequentially.
The method of generating the content is challenged herein as these models collect data from the web and modify it according to the prompt given and gives the final result which originality is contested and also the question of who owns the copyright, the developer of the AI or the person who used his creativity and input a detailed prompt through his creativity and mentions the smallest details which generated a masterpiece. Mirage of Copyright of Generative AI While the Generative AI is being developed every day in this fast-moving world, meanwhile, the laws are getting older; a person can claim copyright, i. e.
, the Indian Copyright Act, 1957, in which section 13(a)4 refers to copyright existing on “original literary, dramatic, musical and artistic works,” the section contradicts with the eligibility of copyright on AI generate outputs for not providing the information on who will hold the copyright over these images, the developer of the AIs who have crafted and gathered the massive data through which the image was generated in no time or the end user who mastered the prompting and crafted a prompt with is intellect, creativity, and engagement mentioning every single details of the image which in result gave an output which is magnificent and feels real. Here, the claims don’t stop; there are in total five stakeholders that can claim the copyright over the generated piece of art. Let's examine them one by one:
2 S. Vishnu Shankar, S. R. Naffees Gowsar & M. Manjubala, Variational Autoencoders (VAEs), in
REVOLUTION WITH GENERATIVE AI: TRENDS AND TECHNIQUES (N. Gayathri, S. Rakesh Kumar & A. Rocha eds., 2025).
3 H. Alqahtani, M. Kavakli-Thorne & G. Kumar, Applications of Generative Adversarial Networks (GANs): An Updated Review, 28 ARCH. COMPUT. METHODS ENG. 525 (2021).
4 The Copyright Act, 1957, § 13(a) (India).
1) The Developer of AI- The developer is the person behind this world of AI.
Companies like Google, OpenAI, and Midjourney invest immense resource on these people to develop the AI and train them and give them a neural architecture. They contend their intellect and give these AI an expression by curating datasets and designing model parameters. These people also hold the protectable expression for the masterpiece generated using these AI’s.
2) The End User/ Prompt Master- The AI is useless in the absence of a master
prompt. The individual who crafts elaborate, thinks and rethinks their ideas and reduce it into a master prompt whose intellect generated the said idea, the said colour theme of graphics, the master tunes of music. The human input reflects meaningful authorial control over the final output; copyright may vest in the prompter as the author of an AI-generated or assisted work.
3) The Original Data creator- Artists, journalists, photographers, and musicians
whose copyrighted works formed the training corpus represent the silent third claim because, in the absence of these datasets, the developer will not be able to give AI a neural architecture. Their rights are infringed at the point of ingestion, even if the generated output is not a verbatim copy of the work.
4) The AI System- The fourth claims come from the AI engine itself, although it is
not legally tenable in any present jurisdiction, AI cannot hold copyright of its generated work. Globally, including India, their exist no authority that recognizes non-human authorship without substantial human creative control.
5) The Employer/ Commissioning Party- The fifth claim is by the employer of the
company, where AI- generated work is produced within an employment relationship, the employer may claim copyright of the work as the ‘work for hire’ doctrine, though this remains legally unsettled in India. Leading Case: India’s Legal Reckoning
1. The RAGHAV Case
The Suryast episode is India's most instructive precedent to date. Using his AI program RAGHAV (“Robust Artificially Intelligent Graphics and Art Visualizer”), artist and lawyer Ankit Sahni reinterpreted his digital sunset photo using the aesthetic framework of Vincent van Gogh's The Starry Night. The Indian Copyright Office denied Sahni's application for sole-author registration, citing RAGHAV as the author, because authorship can only be held by humans. In November 2020, an application that listed Sahni and RAGHAV as co-authors was provisionally registered, potentially making India the first country in the world to acknowledge an AI as a co-author of a work protected by copyright. 5 The Copyright Office sent a withdrawal notice in July 2021, admitting that the registration had been given by mistake and asking Sahni to explain RAGHAV's legal status.
6 The registration is still officially "registered" on the Copyright Office portal, but the issue is still not settled in court. This institutional uncertainty, granting, then taking away, then delaying, sums up India's bigger legal issue: the law doesn't allow AI authorship and doesn't provide a clear alternative framework for AI-assisted works. 7
2. ANI Media Pvt Ltd v Open AI Inc & Anr CS(COMM) 1028/2024
The most consequential ongoing litigation is the copyright infringement suit filed by Asian News International (‘ANI’) before the High Court of Delhi (Justice Amit Bansal) in November 2024. 8 ANI alleged that OpenAI’s ChatGPT was trained on its proprietary and paywalled news content without authorisation, that ChatGPT reproduced ANI’s articles verbatim in user responses, and that the model generated false attributions, damaging ANI’s reputation, the so-called ‘hallucination’ problem. ANI sought ₹2 crore in damages and a permanent injunction. 9 On 19 November 2024, the Court framed four issues for adjudication:10 (i) whether storing ANI’s copyrighted data to train ChatGPT constitutes 5 Indian Copyright Office, Copyright Registration No. A-130396/2020, Suryast by Ankit Sahni (Nov. 2020) (India). 6 S.
Sarkar, Exclusive: Indian Copyright Office Issues Withdrawal Notice to AI Co-Author, MANAGING IP (Dec. 2021), https://www. managingip. com (accessed Mar. 10, 2026). 7 Ankit Sahni’s AI “Co-Authored” Artwork Denied Registration by US, Continues to be Registered in India, SPICYIP (Jan. 3, 2024), https://spicyip. com (accessed Mar. 10, 2026) 8 ANI Media Pvt. Ltd. v. OpenAI Inc. & Anr. , CS(COMM) 1028/2024, Order dated Nov. 19, 2024 (High Ct. of Delhi) (India). 9 ANI v OpenAI in the Delhi HC: Everything So Far and All That Is at Stake, THE LEAFLET (Mar. 17, 2025), https://theleaflet. in (accessed Mar. 10, 2026).
10 ANI Media Pvt. Ltd., CS(COMM) 1028/2024 (Delhi HC Nov. 19, 2024).
infringement under the Copyright Act 1957; (ii) whether generating user responses using that data constitutes a separate act of infringement; (iii) whether such use qualifies as ‘fair dealing’ under section 52 of the Copyright Act 1957;11 (iv) whether the Delhi High Court has territorial jurisdiction, given that OpenAI’s servers are located in the United States. OpenAI, represented by Senior Advocate Amit Sibal, mounted a two-pronged defence. First, on jurisdiction: since its servers and operations are located outside India, Indian courts lack territorial competence. 12. ANI countered that section 62(2) of the Code of Civil Procedure 1908, read with section 62 of the Copyright Act, grants a plaintiff the additional option to sue where it resides or carries on business, a ground that firmly establishes Delhi as a competent forum, given ANI’s headquarters.
Second, on the merits: OpenAI argued that training LLMs involves ‘non-expressive use’ of data, breaking content into statistical tokens rather than reproducing expression and therefore falls outside the scope of copyright infringement. Two amici curiae were appointed to assist the Court. Professor Arul George Scaria (NLSIU, Bengaluru) observed that ChatGPT engages in both expressive and nonexpressive use, acknowledging that certain responses demonstrated significant textual similarity with ANI’s content. 13 Amicus Adarsh Ramanujan maintained that the existing legislative framework, properly interpreted, obliges commercial AI users to obtain licences for copyrighted works.
In February 2025, the Federation of Indian Publishers, the Digital News Publishers Association, and the Indian Music Industry all intervened as interested parties, underscoring the case’s sweeping implications for every creative sector. The Indian courts have already established, in Super Cassettes Industries v Chintamani
11 The Copyright Act, 1957, § 52 (India).
12 Hearing in Copyright Case Against OpenAI Set for March, TECHPOLICY.PRESS (Feb. 14, 2025),
https://techpolicy. press (accessed Mar. 10, 2026). 13 ANI v OpenAI in the Delhi HC: Everything So Far and All That Is at Stake, THE LEAFLET (Mar. 17, 2025), https://theleaflet. in (accessed Mar. 10, 2026) Rao, that section 52 ‘carefully and exhaustively enlists’ exceptions to infringement and that Parliament deliberately confined exceptions to specific categories of works, a strict-construction approach that leaves little room for judicially expanding fair dealing to cover AI training. 14 Similarly, in Syndicate of the Press of the University of Cambridge v B D Bhandari, the Delhi High Court recognised transformative use as a potential exception only where the purpose was substantially different from the original work’s expressive purpose, a standard that OpenAI may struggle to satisfy given its commercial objectives.
15 The Road Ahead: Reforming India’s Copyright Framework India’s Parliamentary Standing Committee on Commerce had, as early as July 2021, recommended reviewing the Patents Act and the Copyright Act to accommodate AI-generated works, recognizing their importance during the COVID-19 pandemic and beyond16. In March 2024, an advisory on AI content labelling was issued by the Ministry of Electronics and Information Technology (MeITY). After that, in February 2025, the commercial use of copyrighted works without a license constitutes infringement under existing law, as affirmed by the response of parliament. 17 A meaningful legislative agenda must rest on four pillars. What are these tests?
Let's have a look into it; firstly, codification of a significant human input test, which is analogous to the European Union’s proposals to know or determine when a human prompter exercises authorial control that is strong enough to get copyright protection. Second, we should set up AI-specific licensing systems and a Text and Data Mining (TDM) exception for non-commercial research, as the EU does. At the same time, we should make sure that commercial AI developers pay rights-holders. Third, requiring AI-generated outputs to be transparent and have watermarks so that they can be traced and held accountable. Fourth, working toward international harmonization to make sure that Indian creators get the same level of protection in digital 14Super Cassettes Indus. v. Chintamani Rao, (2012) 49 PTC 1 (Del. HC) (India). 15 Syndicate of the Press of the Univ. of Cambridge v. B. D.
Bhandari, (2011) (Div. Bench, High Ct. of Delhi) (India). 16 Parliamentary Standing Comm. on Commerce, Review of the Intellectual Property Rights Regime in India, 161st Report (July 2021) (India). 17 Ministry of Electronics & Information Technology, Advisory on AI Platforms (Mar. 2024) (India);Ministry of Electronics & Information Technology, Written Reply to Parliament (Feb. 7, 2025) (India). markets that cross borders. 18 Conclusion The copyright mirage around AI-generated masterpieces is no longer just a philosophical idea; it is a very real legal issue with big economic and cultural stakes. India is at a very important turning point.
The ANI v OpenAI case in the Delhi High Court, the still-unresolved Suryast case, the basic rules in Eastern Book Company v D B Modak, and the originality doctrine in R G Anand v Delux Films all show that the law was never meant to settle disputes between a human prompter, an AI engine, a developer company, and millions of pieces of art that were scraped. So, the mirage is the false belief that the law is good enough to handle the situation. India must act quickly to protect its creative community and take a strong lead in global AI governance. This means making clear laws, fair pay systems, and working with other countries. There are more and more masterpieces than there are laws to protect them, and every day that the government waits to pass new laws, the mirage gets stronger.
18 Rise of the Machines
AI and Copyright, BAR & BENCH (Mar. 31, 2023), https://www.barandbench.com/columns/rise-of-the-machines-ai-and-copyright (accessed Mar. 10, 2026).