Name of author: Deepali Tilwani  (IV Year B.A. L.L.B (Hons.) student at Institute of Law, Nirma)

Name of author: Kritika Singh (IV Year B.A. L.L.B (Hons.) student at Institute of Law, Nirma)

In 2023, the U.S. Copyright Office rejected an application for copyright protection over a graphic novel, Zarya of the Dawn, because key images were generated using Midjourney, an artificial intelligence tool. While the storyline, character development, and sequencing were authored by a human, the Office concluded that the AI-produced visuals lacked the “human authorship” necessary for copyright. The decision sparked global debate: can law meaningfully separate human creative input from machine-generated output in an era of generative technologies?

This dilemma is not confined to copyright ownership alone. It directly implicates the moral rights framework, particularly in jurisdictions like India, where Section 57 of the Copyright Act, 1957 enshrines the author’s rights of paternity and integrity. These rights were created with the implicit assumption that every work reflected a living author’s personality, creative spirit, and identity. However, generative AI produces outputs without human creativity in the traditional sense, often trained on copyrighted datasets without attribution. What is now required is a recalibration of moral rights: a model that protects both human creativity and heritage, and simultaneously subjects AI-generated works to standards of accountability.

This blog proceeds in five parts. It first traces the foundations of moral rights in Indian law and their grounding in international instruments. It then examines how generative AI disrupts the attribution–integrity nexus at the heart of these rights. The discussion moves next to the limits of conferring moral rights on AI systems themselves, while considering the potential for human–AI collaborations to attract protection. Building on comparative developments from jurisdictions such as France, Japan, and the United States, the blog outlines a possible framework for reforming section 57 of the Copyright Act. Ultimately, it highlights the risks of inaction and concludes by proposing a human-centered recalibration of moral rights in the age of generative technologies.

The Doctrine of Moral Rights in India

India’s engagement with moral rights originates in Article 6bis of the Berne Convention, later incorporated through Section 57 of the Copyright Act, 1957. Unlike the economic bundle of rights, Section 57 articulates an author’s “special rights” that operate independently of ownership. These encompass, first, the right of paternity, which secures the ability to be recognised as the creator, and second, the right of integrity, which provides safeguards against any distortion, mutilation, or other alteration of the work that could injure the author’s honour or reputation.

The Indian judiciary has consistently interpreted Section 57 expansively, giving moral rights a significance well beyond the economic sphere. In Manu Bhandari v Kala Vikas Pictures, the Delhi High Court held the author’s claim against a film adaptation that significantly departed from the original novel, signaling that contractual arrangements cannot legitimise derogations from core moral rights. In Amar Nath Sehgal v. Union of India, the destruction of a mural was held to be a grave violation of integrity, with the court linking such rights to the preservation of cultural heritage itself. Yet, these interpretations presuppose a human author and a tangible, stable “work”, an assumption increasingly destabilised by the advent of generative AI, which defies traditional authorship paradigms and remains unaddressed within India’s current statutory framework.

Generative AI’s Disruption of Moral Rights

Generative AI unsettles the doctrinal core of moral rights by dissolving the conventional, person-centred link between creator and creation. Modern generative models produce outputs without a single conscious agent; their “creativity” derives from statistical patterning learned from vast corpora rather than an individuated expressive impulse. This architecture disrupts paternity: the dignity-based impulse behind attribution falters when authorship could plausibly be claimed by a model owner, a coder, a dataset curator, a prompt engineer, or the user who supplied iterative instructions.

Equally acute are integrity problems. Generative systems routinely remix and re-contextualise source materials, producing facsimiles or provocative re-workings that can deform an author’s intended meaning or public persona at scale. Remedies such as injunctions, retractions, or damages seem ill-suited to address algorithmic distortions that travel through opaque datasets and spread instantly across jurisdictions. International responses have been uneven: the Berne-rooted insistence on human authorship remains influential, while regulatory and administrative bodies weigh disclosure, dataset provenance, and remuneration regimes as pragmatic workarounds. However, India has not yet confronted how Section 57’s human-centric vocabulary applies to algorithmic outputs; absent statutory clarity, moral rights risk practical obsolescence or arbitrary extension.

Moral Rights and Human Agency vs Artificial Intelligence

Moral rights presuppose a rights-bearing author capable of reputation, honour, and intention. Systems that generate content through statistical patterning lack legal personhood and agency; they cannot form or communicate an authorial intention and therefore cannot hold moral rights. The U.S. Copyright Office’s guidance reaffirms this predicate: copyright only protects “human authorship,” and material autonomously produced by a generative system is not registrable. That baseline matters for moral rights because paternity (attribution) and integrity (protection against derogatory treatment) are dignity-inflected controls over expression; without a human author, there is no subject of dignity to protect. The normative upshot is clear: AI, as such, cannot be the bearer of moral rights, even where its outputs are aesthetically persuasive, because law ties those rights to human creative personality, not to computational process.

Yet the same guidance leaves space indeed, a pathway for human-AI collaboration to attract protection where the human’s contribution rises to “substantial” authorship. Where a person exercises creative control that shapes the final expression (for example, iterating prompts with an expressive objective, selecting and arranging system outputs, materially editing, or integrating generated material into a coherent design), U.S. practice allows protection for those human contributions while excluding the purely machine-generated portions; applicants must also disclose the presence and scope of AI-generated material. Transposed to India, this maps neatly onto Sec. 57: paternity would attach to the identifiable human contribution (not to the model), and integrity would constrain derogatory treatment of that human-authored expression within a hybrid work. An Indian registration and disclosure protocol requiring provenance logs, prompt histories, and versioned edits would both evidence authorship and cabin claims, thereby reducing disputes over who is the “author” of what in composite outputs.

Possible Framework for India

A measured recalibration of Section 57 can therefore preserve the dignitary core of moral rights while harnessing AI’s productive gains. Statutorily, India could (i) define “author” for hybrid works as the person who makes creative choices that determine the work’s final expressive contours; (ii) mandate disclosure of AI-generated components at deposit and in public metadata; and (iii) recognise a limited “integrity-in-context” claim allowing human authors to object where platform-level remixing or re-contextualisation erases or misattributes their contribution. This approach preserves the Berne-rooted dignitary function of moral rights, supplies the market with auditable authorship signals, and meets a global need for trustworthy AI-era attribution, without diluting the axiom that only people create in the legal sense.

This area of conflict is not new to the world; copyright laws are being tested by the rise of generative AI around the world. For instance, in France, this challenge was recognised and resolved in September 2023 by amending their existing copyright laws to include authorization for AI integration of existing works, mandate labelling as “AI-generated”, etc. Unlike in Japan, where they use a lenient stance to resolve the issue in which the copyright works may be used freely, including for AI training, however, the government has drafted a bill to authorize AI-related ambiguity due to backlashes faced by huge companies against their leniency.

The approaches made by other countries can help India choose its path for such authorization by amending the existing laws to authorize AI-related work, including perpetual and inalienable moral rights (as in France), and other related legislation. The judiciary could also assist and advise on interpreting non-exhaustive terms in the laws, like the interpretation of “authors” given under Section 2(d) of the Copyrights Act, 1957. India can protect the moral rights of its creators in the age of AI by using a balanced framework that includes changes to laws, court decisions, and best practices from around the world.

 Critical Risks of Inaction

The laws have to evolve in parallel with the evolution of new technologies to foster new opportunities and creativity, which could highly impact the country economically and legally. If India fails to address the generative-AI moral rights nexus, it could hinder the creation of an intellectual monoculture, which could lead to disincentivizing new outputs. AI-generated work can easily be detached from contextual sensitivity, which could distort the original work and its interpretation. Lack of legal certainty on authorship and ownership discourages investment in AI creativity and has a chilling impact on innovation, weakening the ability to think critically, criticize, and innovate. Seeing through the lens of the global community, the countries adapting their IP laws to AI-generative work could easily set the IP standards, and India would fall behind other such countries. Without conclusive reform, we risk losing not only economic competitiveness but also the authenticity of our cultural integrity.

Conclusion

Section 57 of the Indian Copyright Act was drafted in an era when authorship was synonymous with individual creativity. Generative AI destabilises this foundation, but the response cannot be to either abandon moral rights or to artificially extend them to machines. A principled middle ground is necessary: moral rights should attach only where human agency contributes original and substantial creative input, supported by verifiable provenance evidence. By codifying such a threshold, India would preserve the dignitary core of Section 57, safeguard cultural and economic interests, and align its copyright system with emerging global standards. In doing so, Indian law would reaffirm that technology may augment creativity, but it cannot replace the author as the bearer of moral rights.