By Sujit Bhar
The rapid rise of generative artificial intelligence has unsettled copyright regimes across the world, and India is now formally grappling with the legal, economic, and philosophical questions it raises. This was made clear in a recent written reply in Parliament by Union Minister of State for Commerce and Industry Jitin Prasada to a question posed by Congress MP Shashi Tharoor. The minister confirmed that the central government is actively reviewing whether the Copyright Act, 1957 adequate to address the challenges posed by generative AI, particularly in relation to authorship, ownership, and “copyrightability” of AI-generated works.
According to the minister’s response, these issues are currently under examination and are expected to be addressed in a forthcoming white paper. He also revealed that the Department for Promotion of Industry and Internal Trade (DPIIT) has constituted an eight-member expert committee, set up on April 28, 2025, to study the implications of generative AI on copyright law. The committee has already completed Part 1 of a working paper focusing on the use of copyrighted content in the training of AI systems. This working paper has been published and opened for stakeholder feedback.
The committee’s mandate includes analysing legal and policy challenges arising from AI’s use within the copyright framework, examining whether existing statutory provisions are sufficient, and making recommendations where changes may be necessary.
While this institutional response is significant, the questions before the committee go far beyond technical amendments to an old statute. At stake are the foundations of how creativity, labour, and value are understood in a world where machines can ingest vast libraries of human expression and generate outputs that often appear indistinguishable from original human work.
TRAINING AI ON COPYRIGHTED MATERIAL
Perhaps the most contentious issue in the AI-copyright debate worldwide is the training of AI systems using copyrighted material available on the internet and other sources. Generative AI models—whether text, image, music, or video-based—are trained on enormous datasets that typically include books, newspaper articles, academic papers, photographs, paintings, and audiovisual works. Much of this material is protected by copyright, yet it is often scraped, copied, and processed without the explicit consent of rights holders.
This has triggered a growing global movement among authors, journalists, artists, photographers, publishers, and their representative organisations. They argue that AI companies are effectively free-riding on decades of human creativity, using copyrighted works as raw material to build profitable products without compensation. Lawsuits against major AI developers in the United States, Europe and elsewhere reflect this backlash, as do collective bargaining efforts by media houses and creators’ unions.
The parallels with earlier battles involving digital platforms are striking. In several parts of the world, Google was eventually forced—through legislation or regulatory pressure—to share revenue with news publishers for displaying snippets or links to their content in search results. The argument then, as now, was that platforms monetised content created by others while returning little or no value to the original creators. Many creators now believe that generative AI poses an even more serious threat because it does not merely link to or summarise content, but absorbs it into systems that can reproduce style, structure, and substance at scale.
There is a real fear that if authors and publishers worldwide demand licensing fees or revenue sharing for every work used in AI training, the economics of AI development could fundamentally change. Training large models is already enormously expensive; adding comprehensive licensing obligations could make it prohibitively so, especially for smaller firms and researchers. In an extreme scenario, this could slow or even halt the development of new AI products, consolidating power in the hands of a few corporations that can afford large licensing deals.
India, as both a major producer of creative content and an emerging AI hub, sits at the heart of this tension. Any approach it adopts will need to balance the protection of creators with the need to foster innovation and technological growth.
CAN AI OUTPUTS BE COPYRIGHTED AT ALL?
A second, equally complex question concerns the copyright status of AI-generated works themselves. If an AI system is trained on existing copyrighted material and then produces text, images, or music by statistically predicting patterns, can the resulting output ever qualify as an “original work” under copyright law?
Traditional copyright doctrine is built around human authorship. Originality, in most jurisdictions including India, is tied to human skill, labour, and judgment. If an AI system merely rearranges, recombines, or rephrases existing material, critics argue that its outputs lack true originality and should not be protected by copyright at all. Granting copyright to such works, they warn, would allow companies or users to monopolise content that is ultimately derived from the collective cultural commons.
On the other hand, denying copyright protection to AI-generated outputs creates its own problems. Businesses that invest heavily in AI-generated content—whether marketing material, design, or entertainment—may be reluctant to do so if competitors can freely copy and reuse the results. This could discourage innovation and commercial deployment of AI tools.
The problem becomes even more acute when one tries to distinguish between human-created and AI-created works. In practice, many outputs are hybrid: a human provides prompts, selects from multiple AI-generated options, edits the output, and curates the final result. At what point does this human involvement cross the threshold into authorship? And how can courts or copyright offices reliably determine whether a work is sufficiently human-made to qualify for protection?
These questions strike at the heart of copyright enforcement. If AI-generated material floods the market, how will rights holders prove that their original work has been copied, as opposed to merely “influencing” an AI model during training? Conversely, how will creators defend themselves against accusations of infringement when AI tools they used may have drawn on copyrighted sources invisibly?
THE LARGER PICTURE
The copyright challenge posed by generative AI is not confined to words. AI-generated visual art, illustrations, photographs, music, and even video are already competing with human-created works in commercial markets. Image-generation models can produce paintings in the style of famous artists, while music models can compose songs that closely resemble existing genres or performers.
These developments raise fresh legal questions. Visual art and music have long been central to copyright law, but AI blurs the line between inspiration and imitation. If an AI generates a painting “in the style of” a particular artist, is that a lawful form of influence or an act of infringement? If such a work is sold, exhibited, or minted as a non-fungible token (NFT), who owns the rights—the user who typed the prompt, the company that built the model, or no one at all?
In India, these issues intersect with a rapidly growing digital art and NFT ecosystem. Indian artists have increasingly used NFTs to monetise their work globally, and AI-generated art could easily enter this space. Without legal clarity, disputes over ownership and authenticity are almost inevitable. The existing Copyright Act, drafted decades before digital art or blockchain technology, offers little guidance on these questions.
Moreover, if AI-generated artworks are granted copyright protection, it could crowd out human artists, especially emerging ones, by flooding the market with inexpensive, machine-generated content. Conversely, if such works are denied protection entirely, the NFT and digital art markets may face uncertainty, undermining investor and creator confidence.
INDIA’S MOMENT OF CHOICE
The DPIIT expert committee’s work, including its published working paper on the use of copyrighted content in AI training, represents an important first step in confronting these dilemmas. The forthcoming white paper is likely to shape the contours of India’s response—whether through statutory amendments, interpretative guidelines, or a combination of both.
India’s challenge will be to craft a framework that protects authors, journalists, and artists from uncompensated exploitation while avoiding an overly restrictive regime that stifles AI research and development. Solutions could include limited exceptions for AI training, collective licensing mechanisms, transparency obligations for AI developers, or new categories of rights tailored specifically to AI-generated works.
What is clear is that the Copyright Act, 1957 now stands at a crossroads. Generative AI forces lawmakers to rethink long-held assumptions about creativity, originality, and ownership. How India resolves these questions will not only shape its own creative and technological future, but may also influence global debates in an increasingly interconnected digital world.
The post A Right to Copy Copycats appeared first on India Legal.