AI tools now write, draw, compose and design at commercial scale. Indian copyright law was written for human authors. The gap between the two is where the next decade of disputes lives.
A small business commissions a logo generated by an AI tool. A marketing team publishes campaign copy the software drafted. A music producer releases a track built on AI-generated stems. Each of them assumes, usually without thinking about it, that they own what they are using. Under Indian law, that assumption deserves more scrutiny than it gets.
The Human at the Centre of Copyright
The Copyright Act, 1957 was built around a human author. Ownership, duration, and moral rights all flow from a person who created the work. The Act does contain an early attempt at handling machines: for computer-generated works, Section 2(d)(vi) defines the author as "the person who causes the work to be created." That language, drafted in 1994, long before generative AI, leaves the central modern question open. When a user types a short prompt and a model produces an image, is the user the person who "caused" the creation in any meaningful authorial sense, or did the system's training and design do the creative work?
Indian courts have not yet settled this, and the Copyright Office's practice has been cautious about registering works claiming an AI as author or co-author. The practical position for now: purely machine-generated output with minimal human creative contribution sits on uncertain ground, while works where a human's skill and judgment demonstrably shaped the result stand on firmer footing. The more you can show of your own selection, arrangement, editing and creative direction, the stronger your claim.
The Training Question, and Why Japan Keeps Coming Up
The second, larger battle is about inputs rather than outputs: whether training AI models on copyrighted material without permission is infringement. India's Copyright Act contains a fair dealing framework under Section 52 that is narrower than the open-ended fair use standard in the United States, and it was not written with machine learning in mind. The question is squarely before the Delhi High Court in ongoing litigation brought by a news agency against a major AI developer over the use of its content in training, a case the entire industry is watching because it will shape what "permission" means at dataset scale in India.
Comparisons with Japan appear constantly in this debate, for a reason. Japan's copyright law contains a notably permissive exception allowing use of works for information analysis, which has been widely read as giving machine-learning training broad room, subject to limits where the use unreasonably prejudices the copyright owner's interests. Whether India should move toward that model, adopt a licensing-based approach, or hold its narrower line is now an active policy conversation involving government consultation with stakeholders. Where India lands will decide whether it competes for AI development on permissive terms or positions itself as a jurisdiction where creators' consent carries more weight.
What Businesses Should Do While the Law Catches Up
Uncertainty is not a reason for paralysis; it is a reason for hygiene. Read the terms of the AI tools you use commercially: providers differ on what rights they grant in outputs, what they claim over your inputs, and what indemnities, if any, they offer. Keep records of the human creative contribution to important assets, prompts, iterations, edits, so authorship can be demonstrated if challenged. Be careful about feeding confidential or third-party copyrighted material into tools whose training and retention practices you have not checked. And for assets that matter, a logo, a core brand work, a flagship product design, consider whether meaningful human authorship, documented, is worth the extra effort compared to accepting whatever the machine produced first.
The Honest Summary
Nobody can tell you today, with certainty, who owns a purely AI-generated work in India, because the law has not decided. What can be said with confidence is that human creative contribution is the safest anchor for ownership, that the training-data question is being fought in an Indian courtroom right now, and that the businesses least likely to be hurt by the eventual answers are the ones keeping records and reading terms while everyone else assumes.