Centre Proposes Mandatory Blanket Licence for AI Training on Copyrighted Works, Royalties for Creators
The Centre has proposed a mandatory blanket licence that would allow artificial intelligence developers to train their systems on all lawfully accessed copyrighted works as a matter of right, while requiring them to pay statutory royalties to copyright owners, rejecting both a broad “fair dealing” expansion and a commercial text and data mining exception. The proposal is contained in...
The Centre has proposed a mandatory blanket licence that would allow artificial intelligence developers to train their systems on all lawfully accessed copyrighted works as a matter of right, while requiring them to pay statutory royalties to copyright owners, rejecting both a broad “fair dealing” expansion and a commercial text and data mining exception.
The proposal is contained in a working paper released by the Department for Promotion of Industry and Internal Trade (DPIIT) on December 8, 2025, titled “Working Paper on Generative AI and Copyright – Part 1: One Nation, One Licence, One Payment: Balancing AI Innovation and Copyright”. The paper invites public comments within 30 days of its publication.
The paper reflects the recommendations of an eight-member committee constituted by DPIIT on April 28, 2025, to examine whether India's copyright law is equipped to deal with the use of protected works in training generative AI systems.
At the core of the proposal is a “hybrid model”, under which AI developers would be permitted to use copyrighted works for training without negotiating individual licences, while creators would be compensated through a statutory remuneration right.
“With a majority view, the Committee decided to recommend a mandatory blanket license in favour of AI Developers for the use of all lawfully accessed copyright-protected works in the training of AI Systems, accompanied by a statutory remuneration right for the copyright holders,” the paper states.
The committee expressly rejected a blanket text and data mining (TDM) exception for commercial AI use, observing that such an approach would “would undermine copyright and it would leave human creators powerless to seek compensation for use of their works in AI Training”.
It said this would not be “a wise policy choice, especially for a country like India which has a rich cultural heritage and a growing content industry with immense potential”.
Opt-out models, under which right holders would be required to actively object to the use of their works, were also rejected. The committee found that such models would disadvantage small creators who often lack awareness, bargaining power, or the means to detect unauthorised scraping.
During stakeholder consultations, the committee found a sharp divide between the technology sector and the creative industry. Most AI and technology companies supported a blanket TDM exception, while content creators and rights holders unanimously favoured licensing-based solutions.
Voluntary licensing was ultimately rejected as impractical, given the scale of the internet and the vast number of copyright owners involved, which would make individual negotiations costly and unworkable.
To operationalise the proposed framework, the paper recommends the creation of a single non-profit collecting body, to be designated by the Central government, called the Copyright Royalties Collective for AI Training (CRCAT).
The body would collect royalties from AI developers and distribute them to right holders through copyright societies and collective management organisations. Non-members would also be eligible to receive royalties upon registration of their works.
The paper clarifies that lawful access remains a prerequisite. AI developers would not be permitted to bypass paywalls or technological protection measures. However, once content is lawfully accessed, no separate permission would be required for its use in AI training under the proposed licence.
To ensure transparency, AI developers would be required to submit disclosures summarising the categories, sources, and nature of training data used. These disclosures would form the basis for apportioning the royalty pool across different classes of works.
Significantly, the working paper also proposes that royalty obligations apply to AI systems that are already commercially deployed and generating revenue, as a corrective measure to ensure fairness and equal treatment across the industry.
It also cautions that continued training of AI systems on unpaid human-created content could, over time, erode incentives for original creation and ultimately harm the AI ecosystem itself. It said the proposed framework seeks to support innovation and human creativity together, by ensuring fair compensation while enabling access to diverse data necessary for developing unbiased and inclusive AI systems.