AI–copyright framework
Why in news?
The DPIIT-led committee has proposed a new AI–copyright framework under which AI companies using copyrighted works for training must pay royalties to creators at rates set by a government-appointed panel.
What the committee has recommended
- A mandatory “One Nation, One Licence, One Payment” blanket licence that allows AI developers to use all lawfully accessed copyrighted works for training, in return for paying statutory royalties to rightsholders.​
- A hybrid model: no individual negotiations at the training stage, but royalties become payable when the AI system is commercially exploited.​
- Royalties to be calculated as a fixed percentage of the gross global revenue (excluding taxes) earned from commercialising AI systems trained on copyrighted content, with the obligation applying even to past training that is already generating revenue.​
Institutional mechanism
- Creation of a central non-profit “Copyright Royalties Collective for AI Training” (CRCAT) to collect royalty payments from AI firms and distribute them to creators, including those outside existing copyright societies who register to receive AI-related royalties.​
- A government-constituted rate‑setting committee comprising senior officials, legal experts, economists and technology experts, plus representatives of CRCAT and AI developers, to periodically fix and review royalty rates; its decisions would be subject to judicial review by courts.​
Rationale and current status
- The committee has explicitly rejected a “zero‑price” or free text‑and‑data‑mining exception, arguing that unrestricted free use of creative works for AI training would harm the creative ecosystem and especially smaller creators.​
- DPIIT has released Part I of the working paper on generative AI and copyright and opened a 30‑day window for stakeholders to send comments, after which the government will decide on legislative or regulatory changes.​
Download Pdf
