Impact Of AI-Generated Works On Indian Copyright Law – Authorship And Originality Challenges

Author: LawSphereX

Category: Intellectual Property

Published: 2026-08-23T20:16:20.379129+00:00

Updated: 2026-08-26T17:11:15.553+00:00

IMPACT OF AI-GENERATED WORKS ON INDIAN

COPYRIGHT LAW – AUTHORSHIP AND ORIGINALITY

CHALLENGES

Shantanu Dipak Jagtap, LL.M., Shivaji University Kolhapur

ABSTRACT

The rapid evolution of artificial intelligence (AI) poses unprecedented challenges for copyright law in India, particularly concerning authorship and originality. Indian copyright law, rooted in the Copyright Act, 1957, presumes human authorship and grants protection only to “original works” created by natural or juristic persons. While Section 2(d)(vi) recognizes “computer-generated works” by attributing authorship to “the person who causes the work to be created,” it was drafted decades before modern generative AI. This creates a legal vacuum for AI-generated works produced with minimal human intervention. The Indian originality standard, articulated in Eastern Book Company v. D. B. Modak, requires “skill and judgment” by the author — a test that purely machine-generated content may fail.

AI-assisted works, by contrast, where human creative input remains significant, are more likely to satisfy both authorship and originality criteria. However, registration controversies, such as the Suryast case involving AI co-authorship, reveal the absence of coherent policy. International approaches vary — the UK recognizes the arranger of a computer-generated work, while U. S. law demands substantial human authorship. Indian policymakers face critical choices: whether to exclude AI-generated works from copyright entirely, adapt current definitions to include them, or create sui generis protections. This paper examines doctrinal gaps, evaluates comparative frameworks, and proposes legislative reforms, including clarifying authorship rules, introducing limited rights for AI-generated outputs, regulating training data usage, and issuing administrative guidance.

Balancing innovation with creative protection will determine whether Indian copyright law can adapt to the realities of AI-driven creation without undermining its human-centered foundation.

Introduction Artificial intelligence is no longer confined to industrial automation or data analysis; it now generates paintings, music, literature, and even legal documents. These developments compel legal systems worldwide to reconsider foundational copyright principles. In India, the Copyright Act, 1957, reflects a human-centric vision of authorship. Section 2(d) defines the author as “the person who creates the work,” with a 1995 amendment introducing Section 2(d)(vi) to address “computer-generated” works. Yet, this provision presupposes human involvement, failing to address autonomous AI creativity. A key threshold for protection is originality, as required by Section 13. In Eastern Book

Company v. D. B. Modak, the Supreme Court adopted the “skill and judgment” standard — rejecting both the minimal “sweat of the brow” test and overly strict creativity thresholds. This test implies active intellectual effort by a human author, raising questions about whether works generated solely by AI qualify. Distinguishing between AI-generated works (minimal human input) and AI-assisted works (significant human creative contribution) becomes crucial in applying these standards. India’s legal framework is further complicated by the absence of authoritative case law or administrative guidance on AI authorship. The Suryast incident — where the Copyright Office briefly recognized an AI as co-author before reconsidering — underscores the uncertainty. Internationally, divergent approaches exist: the UK attributes authorship to the arranger, while U. S. law requires human creativity.

This paper situates the Indian debate within this global context, highlighting doctrinal gaps and exploring policy reforms to address authorship, originality, ownership, and infringement liability in the age of AI. AI-Generated vs. AI-Assisted Works. The first step is to distinguish works autonomously created by AI from those where AI merely aids a human creator. In legal commentary, AIgenerated works are generally defined as those “created autonomously by AI without creative contribution from humans”. 1 By contrast, AI-assisted works involve substantial human creative input, with the AI functioning as a tool or “assistive” element. For example, one analysis notes that AI-assisted works involve “human intervention, and exercise of human creativity (mostly, in the form of programming the AI)”2. In short, AI-generated content is produced by an

1 WIPO, “Artificial Intelligence and Intellectual Property Policy” (2020), https://www.wipo.int

2 NITI Aayog, “Responsible AI for All” (2021), https://www.niti.gov.in

algorithm with minimal human direction, while AI-assisted creations reflect a human author’s choices using AI. This distinction is critical because Indian copyright protection traditionally requires human authorship – a point we address below. Authorship under the Copyright Act, 1957. The Indian Copyright Act presumes authorship by a human person. Section 2(d) generally defines the “author” of a literary, dramatic, musical or artistic work as “the person who creates it”3. Importantly, the Act was amended in 1995 to add Section 2(d)(vi), which treats “computer-generated” works specially: “the author is ‘the person who causes the work to be created’”4. In practice, this means copyright in a computergenerated work vests in the human entity responsible for initiating the process. The Act does not separately define “computer-generated work” – it must be understood by its plain meaning.

In effect, Indian law ties authorship to humans. Commentators note that the Act “unequivocally regards authorship as the qualification of natural persons,” defining an author as an individual “who creates the work” (supra note 3). There is no precedent or guidance allowing a machine to be named an author. Indeed, a leading law firm observes that, even though Section 2(d)(vi) makes a “person” the author of a computer-generated work, that term is ordinarily limited to a natural (or sometimes juristic) person5. Under current law, AI cannot itself hold copyright. At best, copyright in an AI-generated work would vest in the human developer, user or proprietor who “caused” the AI’s output (supra note 4). Because the Act does not define “person” to include a machine, the statute effectively presumes human authorship. For example, Section

22 prescribes a copyright term of 60 years after the author's death, clearly presuming a natural

person as author6. Likewise, Section 51 authorizes infringement suits only against a “person”— again implying a human defendant. The law contains no express recognition of AI as independent author, and practitioners acknowledge that the current “legal framework may not effectively deal with works where the actual creator is not a human”7. Originality Standard in India. Even setting aside authorship issues, any work must meet the originality threshold to be copyrightable. Section 13 of the Act requires an “original” work. The Supreme Court in Eastern Book Company v. D.B. Modak rejected both pure “sweat of the

3 The Copyright Act, No. 14 of 1957, § 2(d) (India).

4 The Copyright Act, No. 14 of 1957, § 2(d)(vi) (India)

5 Artificial Intelligence and Copyright Law in India” (Cyril Amarchand Mangaldas, 2023),

https://www.cyrilshroff.com

6 The Copyright Act, No. 14 of 1957, § 22 (India).

7 FICCI, “Position Paper on Artificial Intelligence and IPR” (2022), https://ficci.in

brow” and overly stringent creativity tests, opting instead for a “skill and judgment” standard8. The Court held the work must reflect the author’s independent skill and judgment rather than mere mechanical effort, requiring “some substantive variation and not merely a trivial variation”. In practice, even modest editorial choices or formatting can satisfy this low creativity threshold. Eastern Book Co. emphasizes originality is judged on intellectual contribution, not labor alone (supra note 9). How do these standards bear on AI-generated works? Purely machine-generated content, without human ingenuity, arguably lacks the required “skill and judgment. ” Indian commentators question whether AI can produce anything truly original; if an AI merely “processes and re-arranges existing data” without human thought, it cannot qualify as original9.

Thus, even if AI-generated content were recognized as a “work” and had a statutory author, it might still fail the originality test under Eastern Book Co. unless significantly shaped by human creative choices. Gaps and Challenges – Copyrightability of AI-Generated Content. Statutory and doctrinal requirements expose gaps when applied to AI outputs. First, the Act’s concept of authorship does not anticipate non-human creativity (supra notes 5 & 6). There is no Indian case law awarding copyright when AI works autonomously. Strikingly, the Copyright Office has issued no policy or circular on AI authorship, and the Registrar offers no criteria to distinguish AIgenerated from AI-assisted works10. Second, it's uncertain whether AI-generated output even qualifies as a “work” under Indian law.

Since copyright protects only works of “original intellectual creation” by an author, mechanically or derivatively produced AI output may be uncopyrightable. Courts and offices remain hesitant to attribute authorship to machines: “AI systems do not possess human consciousness and independent judgment ... are very hesitant to attribute authorship to machines”. Third, Section 2(d)(vi)—enacted long before today's generative AI—implicitly anticipated human involvement (supra note 4). Now, foundation models autonomously create complex outputs from minimal prompts. Legal ambiguity persists: if a user enters a simple prompt, are

8 Eastern Book Company v. D.B. Modak, (2008) 1 SCC 1 (India).

9 Vidhi Centre for Legal Policy, “AI and Copyright: The Indian Perspective” (2023), https://vidhilegalpolicy.in

10 DPIIT, “Copyright Framework in the Age of AI” (2024), https://dpiit.gov.in

they the author? Probably not, as generic prompts lack the detail to satisfy Eastern Book Co. 's “skill and judgment”. The AI cannot qualify, leaving minimal-human-input works unprotected. Fourth, training data poses another challenge. Generative AIs are trained on large data scraped from copyrighted sources, and Indian law offers no text-and-data-mining (TDM) exception such as those in other jurisdictions. For instance, ANI has sued OpenAI for using its content to train ChatGPT without permission11. This highlights legal uncertainty around AI training and underscores the need for clear regulations on dataset use and compensation. Recent Developments: Registration and Policy Signals. The uncertainty in law has already surfaced in the Indian Copyright Office’s practice. The headline example is the “Suryast” case.

In 2020, artist-lawyer Ankit Sahni commissioned an AI app called RAGHAV to generate a painting. In the first instance, Sahni filed a copyright application naming RAGHAV (the AI) as the sole author. The Copyright Office rejected that application outright. Undeterred, Sahni submitted a second application listing himself and RAGHAV as co-authors. To the surprise of many, the office approved that application in November 2020. 12 This was widely reported as the first time an Indian copyright registration recognized an AI tool as a co-author of an artwork. However, this registration proved unstable. About a year later, the Copyright Office issued a withdrawal notice to Sahni, querying the legal status of the AI and suggesting that the certificate of registration might have been issued in error.

13 The withdrawal notice indicated that the office had second thoughts, effectively saying that an artistic work must vest in a human “artist. ” Sahni has since argued that a copyright registration, once granted, cannot simply be vacated administratively – a court proceeding would be needed to “rectify” the register. As of late 2023, the registration of Suryast remains on the books (nominally owned by Sahni), but the controversy is unresolved. 14 11 Reuters, “Indian news agency ANI sues OpenAI for unsanctioned content use in AI training” (Nov. 19 2024)

12 The Copyright Act, 1957, §§ 2(d), 13 (India), available at India Code.

Copyright Office (India), E-Register, Nov. 2020, entry for Suryast (Diary No. 13646/2020-CO/A; RoC No. A- 135120/2020). 13 Copyright Office (India), E-Register, Nov. 2020, entry for Suryast (Diary No. 13646/2020-CO/A; RoC No. A- 135120/2020).

14 Indian Copyright Office, Rectification of Register Procedure, Rule 70, Copyright Rules 2013.

Outside India, Sahni’s case has also drawn attention. In December 2023 the U. S. Copyright Office’s Review Board reaffirmed its earlier decision to deny him copyright in Suryast, on the ground that the work lacks sufficient human authorship. 15 Thus, the same AI-assisted painting is accepted under India’s registry (at least for now) but rejected in the U. S. for want of a human author. Other jurisdictions have taken diverse approaches: for example, the UK automatically grants a computer-generated work to the person who made the “arrangements” for its creation,16 whereas the European Court of Justice has emphasized the need for a human creator. 17 But in India, no court has yet decided an AI-authorship dispute, leaving the Suryast administrative episode as a lone (and still-contested) indicator of policy. Policy Recommendations and Legislative Reforms.

To address these challenges, scholars and lawmakers have urged urgent reform. A Parliamentary Standing Committee on Commerce (2021) explicitly recommended a review of IP laws to accommodate AI. 18 The Committee’s report advocated a “separate category of rights for AI and AI-related inventions” and a thorough revision of existing IPR legislation to incorporate emerging AI technologies. Others have similarly called for “revisiting” the Copyright Act to clarify AI issues. 19 Concrete proposals include: • Clarify authorship and ownership. Amend the Act to spell out who is the “author” when AI is involved. For example, the legislation could explicitly treat humans who input creative prompts or edit outputs as the authors of AI-assisted works, while leaving purely AI-generated works outside traditional copyright (or covered by new rules).

Analogous to the UK’s provision, India might define the author of a computergenerated work as the person who made the arrangements necessary for its creation. [5] In any case, the law should resolve whether rights belong to the AI’s developer, the user (who “caused” the output), or both. Clarifying this in statute would prevent the 15 U. S. Copyright Office Review Board, Second Request for Reconsideration for Refusal to Register Suryast (Dec 2023).

16 Copyright, Designs and Patents Act 1988 (UK), s 9(3)

17 Case C-145/10, Painer v. Standard VerlagsGmbH, 2011 E.C.R. I-12533 (ECJ).

18 Parliamentary Standing Committee on Commerce, Review of the Intellectual Property Rights Regime in India (161st Report, 2021).

19 The Copyright Act 1957, s 51.

piecemeal confusion seen in the Suryast case. • Introduce sui generis rights for AI-generated works. Recognizing that purely AIcreated content may merit some protection, one suggestion is to create a new class of intellectual property specially for AI outputs. 20 Such sui generis rights could grant limited-term exclusivity to AI developers or operators, akin to design rights or database rights, without straining the core copyright principle of human creativity. • Licensing and data usage mechanisms. Given the problem of unauthorized training on copyrighted works, reforms could establish a centralized licensing framework for AI training data. For instance, a self-regulatory organization (SRO) could issue blanket licenses or collect royalties from AI developers, similar to music licensing bodies.

21 Aligning with global trends, India could mimic the EU’s upcoming AI Act (2025) provisions on transparency and opt-out rights: under the EU law, creators can declare if their data may not be used to train AI. • Data protection alignment. Any AI-specific reforms should also dovetail with data privacy laws. India’s proposed Digital Personal Data Protection Act (2023) will regulate personal data usage, which intersects with AI training datasets. 22 Policymakers should ensure that copyright exceptions for AI training do not conflict with privacy safeguards. • Registration and enforcement guidelines. The Copyright Office itself should issue formal guidance on AI-related registrations. Clear instructions that copyright requires human authorship, or criteria for when AI-assisted works are registrable, would avoid arbitrary decisions.

Enforcement rules might also clarify liability for AI infringement: for instance, amending Section 51 or related provisions23 to specify that users or developers can be sued if an AI infringes a copyright. In sum, commentators urge legislative clarity to prevent Indian law from lagging. As one analysis warns, if “proactive steps are not taken right away” – such as revising laws to make “authorship, ownership, [and] liability” clear – India risks impeding innovation and undermining creators’ rights. [13] A balanced reform package could recognize human creativity

20 WIPO, Revised Issues Paper on Intellectual Property Policy and Ar;ficial Intelligence (May 2020).

21 European Parliament and Council, Ar;ficial Intelligence Act, COM/2021/206 final.

22 Digital Personal Data Protection Act 2023 (India).

23 Copyright Act 1957, s 51 (Infringement of Copyright).

in AI-assisted works while structuring appropriate rights and obligations for AI-generated outputs. Conclusion AI’s growing creative capacity challenges the foundational assumption of Indian copyright law — that works are products of human intellect. Current statutory provisions, including Section 2(d)(vi), address only human-caused computer-generated works and are ill-suited to autonomous AI outputs. The Eastern Book Company v. D. B. Modak originality standard further complicates matters, as purely machine-generated works may lack the requisite “skill and judgment” attributable to a human mind. The distinction between AI-generated and AI-assisted works provides a potential pathway for applying existing law: AI-assisted works with substantial human creative input can be protected under current doctrine, while AI-generated works may require alternative forms of protection.

Comparative jurisdictions offer contrasting models — from the UK’s arranger-based attribution to the U. S. ’s strict human-authorship requirement — but none perfectly reconcile the autonomy of AI with traditional copyright principles. The Suryast registration controversy reveals the risks of leaving such questions to ad hoc administrative decisions. Without legislative clarity, inconsistent outcomes will undermine both innovation and legal certainty. This research proposes targeted reforms: (1) defining authorship in AI contexts, (2) introducing sui generis rights for AI-generated works, (3) establishing licensing mechanisms for AI training data, and (4) issuing detailed registration and enforcement guidelines. Ultimately, reform should aim for a balanced approach — safeguarding human creativity, ensuring fair use of data, and fostering AI innovation.

Without such proactive measures, India risks both stifling technological growth and eroding the protection available to genuine human authors in an increasingly AI-mediated creative landscape.

REFERENCES

Statutes i. The Copyright Act, (as amended in 1995) ii. The Digital Personal Data Protection Act, 2023 Cases i. Eastern Book Company v. D.B. Modak, (2008) 1 SCC 1 Books i. Bently Lionel, Sherman Brad, Intellectual Property Law (5th edn, Oxford University Press, 2022) ii. Narayanan P., Copyright and Industrial Designs (6th edn, Eastern Law House, 2017) Online Articles i. Sahni Ankit, “The Suryast Case and AI Authorship in Indian Copyright Law” (2023) https://spicyip.com accessed 8 August 2025 ii. WIPO, “Artificial Intelligence and Intellectual Property Policy” (2024) https://www.wipo.int accessed 8 August 2025 iii. UK Intellectual Property Office, “Copyright and Artificial Intelligence” (2021) https://www.gov.uk/ipo accessed 8 August 2025 iv. EU Parliament, “Artificial Intelligence Act – Provisional Agreement” (2025) https://www.europarl.europa.eu accessed 8 August 2025