New Delhi: The Department for Promotion of Industry and Internal Trade (DPIIT) released the first part of a working paper recommending a new licensing regime for artificial-intelligence (AI) developers training models on legally accessed copyrighted material.
The draft, open for public consultation for 30 days, was drafted by an eight-member committee chaired by additional secretary Himani Pande.
Mandatory licence and royalties
The committee proposes that all AI developers be granted a compulsory blanket licence to use any lawfully accessed copyright-protected works for training. Under this licence, creators would receive statutory remuneration—mandatory royalties—whenever their works are used. Under the model, rights holders would not have the option to opt out of licensing.
To handle licensing and payments, the plan calls for the creation of a centralised collecting body, tentatively named Copyright Royalties Collective for AI Training (CRCAT), to be formed by rights holders and designated by the government. This body would collect payments from AI developers and distribute royalties fairly among creators.
A government-appointed committee will set royalty rates—likely on a fixed-percentage basis of AI developers’ commercial revenue. These rates would be reviewed every three years and remain open to judicial challenge.
Why a hybrid license and not a free-use exception?
The working paper reviewed multiple global approaches—from blanket ‘text-and-data-mining’ exceptions to voluntary licensing—and rejected them, arguing they carry serious downsides. A free or zero-cost licence would undercut incentives for human creators to produce new works. Voluntary licensing, by contrast, could lead to protracted negotiations, high transaction costs, and leave smaller creators and start-ups at a disadvantage.
According to the committee, the hybrid licence system strikes a balance. It preserves creators’ rights and incomes while giving AI developers predictable, streamlined access to large training datasets. The centralised structure and single payment mechanism can lower compliance burden, especially for start-ups and small players.
Wider context: Balancing innovation and creative rights
The DPIIT working paper highlights a core tension- AI developers need vast, high-quality data to build robust models; creators need protection and fair remuneration for their original content. Unregulated use of copyrighted works risks undermining human creativity, potentially leading to a decline in original content as AI-generated content becomes easier and cheaper.
At the same time, long licensing negotiations and high costs could stifle innovation—especially for small AI firms. By offering a uniform, statutory licence with agreed royalties, the proposal aims to level the playing field for both large firms and smaller companies working in emerging AI sectors.
What happens next?
The working paper calls for public feedback within 30 days. If the proposed framework is adopted, it could make India one of the first large economies to impose a compulsory licensing model on AI firms for training data—potentially reshaping the economics of AI development and content creation in the country.
Stakeholders, including artists, publishers, and tech firms, will now have a chance to weigh in. The final policy may strike a delicate balance between protecting creative rights and fostering technological innovation.
Also Read
Expanding IT/ITeS Ecosystem Ensures AI-Powered Cybersecurity Measures by West Bengal Govt
