Key takeaways
- The DPIIT working paper proposes a hybrid model with a mandatory blanket licence for AI training and royalties on commercialisation.
- It is a proposal only, and no bill or amendment was confirmed in the sources reviewed.
- ANI v OpenAI in the Delhi High Court may shape the law before any legislation.
- US AI companies should map Indian-origin data and keep provenance records.
India’s AI copyright licence proposal came from a government working paper rather than a statute, but it has drawn attention from US AI developers because it would reshape the rules for training models on Indian content. The Department for Promotion of Industry and Internal Trade released the first part of its working paper on generative AI and copyright on 8 December 2025, and it suggested a hybrid model that includes a mandatory blanket licence for training. The India AI copyright licence idea is still a proposal.
This guide explains what the paper proposes, how it differs from US and other approaches, what is unsettled, and what US AI companies should do now. It is written by an Indian law firm for US readers. It is general information, not legal advice, and it describes a proposal that may change. Check DPIIT’s site for the latest status before relying on any detail.
What the working paper proposes
The DPIIT working paper, prepared by a committee formed in 2025, addresses whether and how AI developers may train on copyrighted works. The DPIIT working paper generative AI part one rejects a zero-price licence, which would let developers use works without paying, arguing that it could reduce the supply of original human-created work. It also treats a pure opt-out approach as problematic. The India AI copyright licence debate began with the DPIIT working paper.
Instead, it proposes a hybrid model built around a mandatory blanket licence for AI training. Under reports of the proposal, developers could train on lawfully accessed works without negotiating with each owner, and pay royalties only when the AI system is commercialised. Rates would be set by a government-appointed committee, subject to judicial review. The mandatory blanket licence AI training model is intended to give certainty to developers and payment to creators.
Royalties would be collected centrally and shared with creators, including those who are not members of copyright societies. Commentary describes a royalty on commercialisation AI India would apply once a system is commercially deployed, rather than at the point of training. The paper was released for public comment for 30 days, and later parts were expected to address other issues such as outputs.
How it differs from other approaches
The US has no statutory AI training licence. Training disputes are being decided under the fair use doctrine in a series of lawsuits, with mixed early results and settlements. The EU has a text and data mining exception with an opt-out for rights holders in its copyright directive and AI Act obligations on transparency. The UK has consulted on its own approach. India’s proposal would be a statutory licensing model, unlike all of these.
| Region | Approach | Status |
|---|---|---|
| India | Mandatory blanket licence with royalties on commercialisation | Working paper, December 2025 |
| United States | Fair use decided case by case | Litigation and settlements |
| EU | Text and data mining exception with opt-out | In force |
| UK | Consultation on approach | Under review |
A statutory licence trades control for certainty. Owners could not refuse training but would be paid. Developers would not need to secure licences from each owner but would pay under a set formula. If adopted, it could also apply to foreign developers whose models are used in India or trained on Indian works, depending on the drafting of the territorial scope.
India also has litigation in progress. ANI v OpenAI in the Delhi High Court, filed in 2024, raises questions on training, storage and output under the Copyright Act, and several publishers and industry bodies have intervened. The ANI v OpenAI Delhi High Court case could shape the law before any legislation. Follow its progress. If adopted, the India AI copyright licence could apply to US AI companies Indian training data uses.
Open questions and criticism
Critics, as in opinion pieces such as one from Deccan Herald, argue that a government-mandated rate-setting body is hard to design, that the system may stifle innovation and that it may not fairly compensate creators if rates are low. There are also concerns that global model developers might avoid Indian data if the terms are unattractive, and about administrative complexity. Critics of the India AI copyright licence worry about rate setting.
Several practical questions remain. What does lawfully accessed mean for scraped content behind terms of service? How would the licence treat foundation models trained abroad but used in India? How would rates reflect the value of different works? How are creators identified and paid? Would the licence cover outputs and the use of retrieval systems? The paper does not settle everything.
Legislative status is also uncertain. The sources reviewed show the paper and commentary, but not a bill or an amendment to the Copyright Act, 1957. A working paper is not law. Timing depends on consultation outcomes and government priorities. Treat the proposal as a signal of direction, not a rule you must follow today.
What US AI companies should do
Start by mapping your exposure. Identify which training datasets include Indian-origin content, which products are offered in India and whether you license data from Indian publishers. Review contracts with data suppliers for warranties on rights. Keep records of data provenance, since any licence regime and litigation will ask for it.
Engage with the process. DPIIT and industry bodies invite comments, and later parts of the working paper may open further consultation. US companies can respond through trade associations or Indian counsel. Consider the economics: how would a royalty on commercialisation affect your pricing in India? Model scenarios.
Stay compliant today. Current Indian law has no AI-specific training exception, and the Copyright Act’s fair dealing provisions are narrow. Avoid assuming that US fair use arguments carry over. For Indian content, consider licences where possible. Seek legal advice on data sourcing and on the Delhi High Court litigation. Prepare now for an India AI copyright licence by tracking data provenance.
Next steps and quick answers
Track DPIIT announcements and the ANI case, map your datasets, review contracts and prepare to comment. This is general information, not legal advice. Halverton & Co. does not advise on US law. Check DPIIT announcements for any India AI copyright licence update.
Related reading: the DPIIT AI copyright working paper, ANI v OpenAI and AI training in India, AI copyright in India and the US compared, AI training data licensing with Indian publishers, scraping Indian websites for AI training and AI hallucination sanctions for lawyers.
Quick answers
Is the AI copyright licence law in India yet?
No. It is a proposal in a DPIIT working paper released in December 2025. Check for later developments.
What does the proposal say about royalties?
Reports say royalties would be due once an AI system is commercialised, at rates set by a government-appointed committee, and collected centrally.
How does it differ from US fair use?
It would be a statutory licence with payment, whereas US training disputes are decided under fair use case by case.
Is the Indian AI copyright licence law in force?
No. It is a proposal in a working paper. Check DPIIT for later developments.
Final word
Halverton & Co. is an Indian law firm advising US companies, founders, investors and law firms on Indian law, including Indian copyright, technology and AI policy advice for US and global AI companies. We practise in Jharkhand, Maharashtra and before the Supreme Court of India, and work alongside US counsel, who advise on US law. Halverton & Co.: Where tech needs law! If you have a question about Indian law, write to us at office@halvertonandco.com, or get in touch.
This article reflects developments reported up to 31 October 2026. It is for general information only, is not legal advice, and does not create an attorney-client relationship. Halverton & Co. is an Indian law firm and does not advise on US law; US-law points should be confirmed with US counsel.
Sources
Related practice area
Technology & AI Law
Contracts, data protection and AI policies for products that handle other people’s data.
View the practice areaComments
No account needed. Comments are reviewed before they appear, so yours will show here once approved. Please do not share confidential details or describe your own legal matter: replies are general discussion, not legal advice, and do not create an advocate-client relationship. For your own situation, write to office@halvertonandco.com.




