CHENNAI: The Department for Promotion of Industries and Internal Trade (PIIT) has released a working paper proposing a sweeping new framework for how artificial-intelligence (AI) developers in India may use copyrighted material to train their models. Under the proposal, AI firms would be required to obtain a mandatory blanket licence for any “lawfully accessed” copyright-protected content they incorporate into their training datasets. Instead of negotiating individual deals with every copyright holder — whether musicians, authors, journalists, or artists — the blanket licence would enable developers to access the full range of legally acquired works with a single licence.
To protect the rights of creators, the framework calls for a statutory remuneration mechanism. Copyright holders would be eligible to receive royalties when their work is used for AI training. These payments would be collected and distributed by a new centralised, non-profit body — formed by rights-holders and designated by the government — intended to serve as an umbrella organisation for licensing and royalty distribution. Royalty rates would be set by a government-appointed committee and remain subject to judicial review. This “one licence, one payment” model would also relieve AI developers — especially start-ups and smaller players — of the burden of negotiating multiple individual licences, thereby lowering compliance costs and legal uncertainty.
The DPIIT committee rejected proposals for a “zero-price licence” — that is, allowing unrestricted use of copyrighted works without compensation. The paper warned that such an approach could undermine incentives for human creators, eventually harming the production of original, high-quality creative content. The blanket licence model, in contrast, aims to strike a balance: enabling rapid development and innovation in AI while ensuring fair compensation and protection for original creators.
The working paper is now out for public consultation for 30 days. Stakeholders of all kinds — from authors, publishers, artists to AI developers and tech firms — are invited to express their views.
What this could mean in practice:
If this proposal becomes policy, it could significantly reshape how AI training is done in India. For AI companies and startups, it may simplify access to large and diverse training datasets, reducing licensing friction and legal risk. For creators — authors, artists, news organisations, musicians — it promises a steady royalty revenue stream every time their work contributes to AI model training. On the flip side, some creators may worry about loss of control over how and where their works are used, since under the proposed licence they would not be able to opt out. Meanwhile, AI developers might face new recurring costs in the form of royalty payments, which could impact business models, especially for smaller firms.
In the bigger picture, this marks a bold attempt by India to chart a middle path between unregulated AI growth and over-protective copyright law. As generative AI becomes more pervasive, such regulation may set a precedent not just domestically but globally — reflecting India’s desire to protect creator rights while nurturing its own AI ecosystem.