The rapid adoption of generative artificial intelligence (AI) is exposing gaps in India’s copyright framework, particularly around authorship, originality, AI-generated works and the use of copyrighted material for training AI models.
Under the Copyright Act, 1957, Section 2(d)(vi) defines the author of a computer-generated work as “the person who causes the work to be created”. While the provision was framed around computer-assisted creation, the emergence of generative AI systems capable of producing complex text, music, images and other creative outputs has complicated the question of where human authorship ends and machine-generated content begins.
Zameer Nathani, Group General Counsel, DNEG, said the existing framework does not clearly distinguish between AI-assisted works, where a human exercises meaningful creative control, and works generated with limited human intervention.
“While this definition sufficed when software served merely as an operational tool, modern generative AI creates complex outputs based on autonomous processing, blurring the boundary between human agency and machine automation,” Nathani said.
According to Nathani, where creators exercise substantial skill and creative judgment — including through refining prompts, selecting and modifying outputs or integrating AI-generated material into a larger original work — copyright protection may be argued under the originality standard laid down by the Supreme Court in Eastern Book Company v. D.B. Modak.
However, where human participation is minimal or largely limited to providing instructions, establishing authorship under Section 2(d) becomes more difficult, he said. Indian copyright law does not recognise a non-human AI system as a legal author.
The issue extends beyond authorship to the use of copyrighted material to train AI models.
Section 52(1)(a) of the Copyright Act provides for specified fair-dealing exceptions, but does not expressly set out a framework for automated data scraping or large-scale AI training. Recent litigation, including proceedings before the Delhi High Court in ANI Media Pvt. Ltd. v. OpenAI OpCo LLC, has brought the application of copyright and fair-dealing principles to commercial AI systems under greater scrutiny.
Nathani said the absence of specific statutory rules leaves significant questions unresolved and called for legislative intervention.
“Parliament must introduce comprehensive legislative updates,” he said, adding that reforms could include statutory thresholds for human intervention in creative works, opt-out mechanisms for rights holders whose works are used for model training, and remuneration frameworks for creators whose original works contribute to AI technologies.
‘Loss of originality and voice’
The legal uncertainty is accompanied by concerns from creators about the impact of AI on the creative process itself.
Saira Shah Halim, author of Comrades and Comeback, said AI can provide useful assistance to independent authors, particularly for tasks such as self-editing, indexing and proofreading that traditionally required external publishing support.
“Authors working independently can leverage AI tools for self-editing, indexing, or proofing — tasks historically requiring external publishing support,” Halim said.
She said this can democratise access to tools and help with time management, but argued that excessive reliance on AI could affect originality and individual voice.
“Over-reliance on AI models risks standardising narrative rhythms, diluting unique authorial voices, and encouraging repetitive or formulaic storytelling,” she said.
Halim also said that the existing copyright framework was not designed for the scale and autonomy of modern generative AI.
“Under current Indian jurisprudence, the Copyright Act, 1957 is fundamentally unprepared to handle modern autonomous Generative AI. While it offers limited scope for computer-assisted human work, it presents significant statutory and judicial gaps,” she said.
Musicians raise concerns over AI training
The concerns are particularly pronounced among some musicians and artists, who have questioned the use of human-created works as training material for generative systems.
Musician and artist Deepak Peace said awareness around the copyright implications of AI remains limited and called for greater protection of human-created music.
“There is very less awareness about it. I don’t think we have any laws but we should urgently have some in place,” Peace said.
He objected to human-created music being used to train machines capable of generating new music, arguing that the creative process itself has value beyond the resulting work.
“You can copy music as a mathematical pattern but music is born from someplace within us that we don’t know,” Peace said.
For Peace, the question is not only whether AI can reproduce elements of music, but whether the creative act can be separated from the human experience behind it.
“The joy is in making — the process of music. I love it. Why would I outsource that joy,” he said.
The growing debate therefore spans two separate but connected questions: whether existing copyright principles can determine when AI-assisted output qualifies for protection, and whether creators should have control or receive compensation when their copyrighted works are used in the development of AI systems.
With generative AI increasingly becoming part of the creative workflow, the distinction between a tool used by an author and a system generating content autonomously is likely to remain a central challenge for copyright law.
