Author: Priyanshu Pareek
College: JECRC UNIVERSITY
Abstract
When an AI system autonomously generates a poem, composes a painting, or produces a musical score, a question arises that India’s Copyright Act, 1957, enacted decades before the internet existed, is constitutionally incapable of answering: who is the author? The Act defines authorship exclusively in terms of human or juristic persons; it offers no framework for machine-generated creativity. In April 2026, the Delhi High Court directed the Copyright Office to decide within eight weeks whether AI-generated art is eligible for copyright protection in India, making this the most pressing unresolved question in Indian intellectual property law. This article examines the statutory void at the intersection of artificial intelligence and copyright, analyses the three-way conflict between copyright claimants in AI-generated works, traces the comparative frameworks emerging in the United Kingdom, United States, European Union, and China, evaluates the December 2025 DPIIT Working Paper’s proposed hybrid licensing model, and argues that India requires an express legislative amendment not administrative guidance to resolve a question that goes to the foundational architecture of intellectual property protection in the digital age.
To the Point
Ankit Sahni, a New Delhi-based lawyer, used an AI tool named RAGHAV to create a painting called ‘Suryast’ a work the Indian Copyright Office registered with RAGHAV listed as co-author alongside Sahni. When the registration attracted public scrutiny, the Copyright Office issued a withdrawal notice, citing Section 2(d) of the Copyright Act, 1957, which limits authorship to natural persons. The application remains listed as ‘registered’ on the Office’s website. The court has yet to decide. That unresolved entry in a government database encapsulates the entire problem: India has copyright registrations it cannot justify, disputes it cannot adjudicate, and legislation that predates the technology it is being asked to govern.
The commercial stakes are not abstract. India’s AI industry is projected to contribute over USD 500 billion to the economy by 2030. Generative AI tools are producing content across film, music, software, journalism, and visual art at industrial scale. The ANI Media v. OpenAI case (2024) India’s first major copyright litigation involving an AI company is pending before the Delhi High Court. The Federation of Indian Publishers, NDTV, Indian Express, and Hindustan Times have all filed or threatened similar actions. In December 2025, the Department for Promotion of Industry and Internal Trade (DPIIT) published India’s first Working Paper on Generative AI and Copyright. The working paper is significant. It is not a statute. The gap between administrative guidance and legislative mandate is precisely where India’s copyright crisis lives.
Use of Legal Jargon
The Copyright Act, 1957, protects original literary, dramatic, musical, and artistic works under Section 13. The term ‘original’ has been interpreted by the Supreme Court in Eastern Book Company v. D.B. Modak (2008) 1 SCC 1 to require a ‘modicum of creativity’ a degree of intellectual effort and judgment that goes beyond mere mechanical reproduction. This ‘skill and judgment’ standard is directly relevant to AI: if the creative choices in an AI-generated work are made by the machine rather than by a human, the work fails the originality test as currently construed by Indian courts.
Section 2(d) of the Act defines ‘author’ across six categories of works. For literary, dramatic, musical, and artistic works, the author is the person who creates the work. For computer-generated works a category introduced by the 1994 amendment Section 2(d)(vi) defines the author as ‘the person who causes the work to be created.’ This provision was designed for software programs, not generative AI. Its application to AI-generated content creates three competing claimants: the AI developer who built the system, the user who provided the prompt, and theoretically the AI itself a claim the current law rejects because AI has no legal personality under Indian law.
Section 51 of the Act defines copyright infringement as any act that falls within the exclusive rights of the copyright owner without authorisation or a statutory exception. Section 52 enumerates fair dealing exceptions, including private use and research. The DPIIT has clarified in its Working Paper that commercial-scale AI training on copyrighted datasets does not qualify for the Section 52 fair dealing exception a position that aligns with Article 13 of the TRIPS Agreement, which permits exceptions to copyright only where they do not conflict with the normal exploitation of the work. This interpretation, if upheld by courts, means that every major AI company training its models on Indian copyrighted content without a licence is in prima facie violation of Section 51 of the Copyright Act.
The concept of moral rights under Section 57 the author’s right to claim authorship and protect the work from distortion is equally disrupted by AI. If no human author can be identified for an AI-generated work, moral rights are extinguished by default. This has practical consequences: an AI-generated painting that is subsequently distorted or misattributed has no legal claimant to enforce the right of integrity under Section 57.
The Proof
The three legal problems posed by generative AI are distinct and require separate analysis. First, the authorship vacuum: when a work is generated entirely by an AI system without meaningful human creative input, it falls outside the protection of the Copyright Act, 1957, because no qualifying author exists. It enters the public domain by default meaning the company that invested resources in building the AI tool receives no intellectual property protection for its commercial output. This creates a perverse incentive structure: the more autonomous and sophisticated the AI, the less legal protection its output enjoys.
Second, the training data problem: large language models and image generation systems are trained on datasets comprising billions of copyrighted works books, articles, photographs, musical recordings, software code scraped from the internet without licence or compensation. In the ANI Media v. OpenAI case (2024 SCC OnLine Del 8120), the Delhi High Court is examining whether OpenAI’s use of ANI’s copyrighted news content to train ChatGPT constitutes infringement under Section 51. In November 2025, the Regional Court of Munich ruled in GEMA v. OpenAI (Case No. 42 O 14139/24) that AI models infringe copyright when they memorise and reproduce protected works without authorisation the first European court ruling on this question and a significant data point for the Delhi High Court in ANI.
Third, the deep-fake and personality rights problem: generative AI can synthesise a deceased artist’s voice, replicate a living musician’s style, or produce photorealistic images of real individuals engaging in events that never occurred. India has no statutory personality rights framework. The DPIIT Working Paper of December 2025 proposes a statutory registry allowing performers and public figures to record non-consent regarding the use of their persona in AI training a proposal that, while necessary, remains unlegislated. The intersection of deepfakes, AI-generated content, and the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, which prohibit ‘morphed images’ of individuals, creates overlapping and uncoordinated regulatory exposure.
Case Laws
Eastern Book Company & Ors. v. D.B. Modak, (2008) 1 SCC 1
The Supreme Court held that for a work to qualify for copyright protection under Section 13 of the Copyright Act, 1957, it must demonstrate a ‘modicum of creativity’ the author must exercise skill, judgment, and intellectual effort beyond mere mechanical copying. This precedent directly governs the AI authorship debate: since AI systems generate output through statistical pattern recognition rather than conscious creative judgment, courts applying this standard would likely hold that purely AI-generated works lack the originality necessary for copyright protection under existing Indian law.
Ankit Sahni / RAGHAV Indian Copyright Office (2020–2021) [Suryast Case]
Ankit Sahni applied for copyright registration of ‘Suryast’, listing himself and AI tool RAGHAV as co-authors. The Indian Copyright Office initially registered the work the first such registration acknowledging AI co-authorship globally. The Office subsequently issued a withdrawal notice, holding that the Copyright Act, 1957, requires authorship to be attributed to natural persons. In April 2026, the Delhi High Court directed the Copyright Office to determine within eight weeks whether AI-generated art can be protected under existing law. This case is the direct constitutional flashpoint for Indian AI-copyright jurisprudence.
ANI Media Private Limited v. OpenAI Inc., 2024 SCC OnLine Del 8120 (Delhi HC — Pending)
India’s first major AI copyright litigation. ANI Media, a leading news agency, alleged that OpenAI’s ChatGPT was trained on and reproduces ANI’s copyrighted content without licence or compensation, constituting infringement under Section 51. The Delhi High Court is examining questions of training data liability, the applicability of Section 52 fair dealing to commercial AI training, and whether Indian courts have jurisdiction over a foreign AI company whose servers are located abroad. No final judgment has been delivered. When it arrives, it will be the most consequential copyright ruling in Indian legal history.
Thaler v. Vidal, 43 F.4th 1207 (Fed. Cir. 2022) USA [Comparative]
The US Court of Appeals for the Federal Circuit held that under the Patent Act, only natural persons can be inventors rejecting Stephen Thaler’s application to list AI system DABUS as the inventor of two patents. The US Copyright Office subsequently confirmed in 2023 that it will not register works produced solely by AI without human authorship. This precedent has been widely cited in Indian academic commentary and is directly relevant to the RAGHAV case before the Delhi HC India is likely to reach the same conclusion through judicial rather than administrative means.
GEMA v. OpenAI, Case No. 42 O 14139/24 (Regional Court of Munich, November 2025) Comparative
Germany’s music rights society GEMA sued OpenAI for using copyrighted song lyrics to train ChatGPT without authorisation. The Munich court ruled in GEMA’s favour, holding that AI models infringe copyright when they memorise and can reproduce protected works verbatim. This is the most recent and directly relevant comparative precedent on the training data liability question pending before the Delhi High Court in ANI v. OpenAI and the first European court to establish that ‘memorisation’ of copyright-protected content within AI model parameters constitutes an act of reproduction under copyright law.
Conclusion
India’s Copyright Act, 1957, was designed for a world where every creative work had a human author. Generative AI has produced a world where the most commercially significant creative outputs may have no human author at all. The RAGHAV registration, the ANI litigation, and the Delhi High Court’s April 2026 directions together constitute a judicial acknowledgment that existing law is insufficient but judicial acknowledgment is not legislative resolution. The DPIIT’s December 2025 Working Paper on Generative AI and Copyright is a welcome first step and a substantive policy document, but it is not law. Administrative guidance cannot amend a statute.
Parliament must act on three specific fronts. First, Section 2(d)(vi) of the Copyright Act, 1957, must be amended to expressly define ‘AI-assisted works’ where human creative input is meaningful as distinct from ‘AI-generated works’ where the AI is the primary creative agent. AI-assisted works should attract standard copyright protection vesting in the human prompter or supervisor. AI-generated works should either be placed in the public domain or attract a new sui generis right vesting in the entity that caused the work to be created. Second, a statutory text and data mining exception modelled on Article 4 of the EU Copyright Directive (2019) must be introduced, permitting AI training on commercially available datasets subject to opt-out mechanisms for rights-holders, and requiring fair compensation through a statutory licensing mechanism for commercial training use. Third, a Personality Rights Protection provision must be inserted, allowing individuals and performers to register non-consent regarding the synthetic reproduction of their voice, likeness, or creative style. The ghost in the machine is writing faster than Parliament can legislate. The question is not whether India’s copyright law needs reform it is whether Parliament will act before the courts are forced to build the framework case by case, precedent by precedent, in the absence of the statute that should have been written already.
FAQ
Q1. Can AI be recognised as an author under Indian copyright law?
No. The Copyright Act, 1957, limits authorship to natural or juristic persons under Section 2(d). AI has no legal personality in India and cannot hold rights. For AI-generated works, the closest applicable provision is Section 2(d)(vi), which designates the ‘person who causes the work to be created’ as the author but this provision was designed for software, not autonomous generative AI.
Q2. Does using copyrighted content to train an AI model amount to infringement in India?
Currently unresolved. The DPIIT’s December 2025 Working Paper confirmed that commercial-scale AI training does not qualify for the Section 52 fair dealing exception. The ANI v. OpenAI case before the Delhi High Court will be the first judicial determination of this question. The Munich court’s November 2025 ruling in GEMA v. OpenAI holding that AI memorisation of copyrighted lyrics constitutes reproduction — provides a persuasive comparative precedent.
Q3. What was the RAGHAV case and what is its current status?
Ankit Sahni applied for copyright registration of ‘Suryast’, an AI-assisted painting, listing AI tool RAGHAV as co-author. The Indian Copyright Office initially registered the work but later issued a withdrawal notice. In April 2026, the Delhi High Court directed the Copyright Office to determine within eight weeks whether AI-generated art is eligible for copyright protection under the existing Act making it the first Indian judicial proceeding to directly address AI authorship.
Q4. What has India’s government done to address AI and copyright?
In May 2025, the Ministry of Commerce constituted an eight-member expert panel to review the Copyright Act, 1957, for AI-related gaps. In December 2025, the DPIIT published a Working Paper on Generative AI and Copyright proposing a hybrid licensing model for AI training data. As of mid-2026, no legislative amendment has been introduced in Parliament. The expert panel’s recommendations and the Delhi HC’s ruling in the RAGHAV matter are expected to form the basis of future reform.
Q5. How do other countries handle copyright in AI-generated works?
The UK grants copyright for computer-generated works to ‘the person who makes the necessary arrangements’ under Section 9(3) CDPA, 1988. The USA requires substantial human authorship and will not register purely AI-generated works (US Copyright Office, 2023). The EU is developing a ‘significant human input’ test under the AI Act framework. China is the most permissive, with courts beginning to protect AI-generated works. India has no express provision and will need to choose between these models through legislative reform.
References
- Copyright Act, 1957 (India), Sections 2(d), 13, 14, 17, 51, 52, 57.
- Eastern Book Company & Ors. v. D.B. Modak, (2008) 1 SCC 1 (Supreme Court of India).
- ANI Media Private Limited v. OpenAI Inc., 2024 SCC OnLine Del 8120 (Delhi High Court — Pending).
- R.G. Anand v. Deluxe Films, (1978) 4 SCC 118 (Supreme Court of India).
- Civic Chandran v. C. Ammini Amma, (1996) Kerala HC four-factor fair use test.
- Thaler v. Vidal, 43 F.4th 1207 (Federal Circuit, USA, 2022).
- GEMA v. OpenAI, Case No. 42 O 14139/24 (Regional Court of Munich, Germany, November 2025).
- Copyright, Designs and Patents Act 1988, Section 9(3) (United Kingdom).
- EU Artificial Intelligence Act, Regulation (EU) 2024/1689; EU Copyright Directive 2019/790, Article 4.
- TRIPS Agreement, 1994, Article 13.
- DPIIT Working Paper on Generative AI and Copyright (December 8, 2025). Available at: https://dpiit.gov.in
- Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021 (India).
- IAM Media, ‘Inside India’s IP Market 2026: The Impact of Generative AI on Copyright as Indian Jurisprudence Evolves’. Available at: https://www.iam-media.com/guide/india-managing-the-ip-lifecycle/2026/article/the-impact-of-generative-ai-copyright-indian-jurisprudence-evolves
- Bar and Bench, ‘India’s Generative AI Moment: Copyright Law at Regulatory Crossroads’ (January 2026). Available at: https://www.barandbench.com/columns/indias-generative-ai-moment-copyright-law-at-regulatory-crossroads
- Lexology / Khurana & Khurana, ‘Generative AI & Copyright Law in India: Who Owns Machine-Made Works?’ (July 2025). Available at: https://www.lexology.com/library/detail.aspx?g=ec2ab57f-9934-45c6-9bce-fd7e6c84db71
- Mondaq / De Penning & De Penning, ‘AI and Copyright Law: Who Owns AI-Generated Content?’ (March 2026). Available at: https://depenning.com/blog/ai-and-copyright-law-who-owns-ai-generated-content/
