AI Training Royalty: India Proposes Mandatory Blanket Licence To Regulate Copyright Use and Guarantee Fair Compensation for Creators

Mumbai, November 9: The Indian government has proposed a major overhaul of how artificial intelligence (AI) companies use copyrighted material to train their models. A new framework outlined in a working paper titled “One Nation, One License, One Payment: Balancing AI Innovation and Copyright” recommends a mandatory blanket licence that would require all AI developers to pay royalties to creators, including musicians, authors, news publishers and other copyright holders.

This proposed system rejects voluntary licensing arrangements, such as individual deals between AI companies and publishers. According to the committee behind the paper, voluntary agreements favour only large corporations and can leave smaller creators and start-ups at a disadvantage. The government argues that a mandatory licensing system would give AI developers equal access to training data while ensuring fair compensation for all copyright owners. New Android Malware Alert! FvncBot, SeedSnatcher and Upgraded ClayRat Can Hack Your Mobile Phone and Steal Banking Data; Here’s How To Protect Your Device.

If implemented, India would become the first country to introduce a statutory licensing regime specifically for AI training. The framework has been prepared by a committee under the Department for Promotion of Industry and Internal Trade (DPIIT) and is currently open for stakeholder feedback.

A key element of the proposal is the creation of a new body called the Copyright Royalties Collective for AI Training (CRCAT). This organisation, made up of rightsholders, would be responsible for collecting and distributing royalties. Only one organisation per category of copyrighted work would be allowed to join, and its membership base may expand over time to include more sectors.

Royalty rates would be set by a government-appointed panel consisting of legal, economic and technical experts, along with representatives from CRCAT and the AI industry. The committee would review and adjust rates every three years. The paper suggests that a flat-rate percentage of global revenue from AI systems trained on copyrighted material may be the most practical model.

The proposed royalty payments would apply retroactively, meaning AI companies already using copyrighted data to train commercial models would need to pay for past usage. Companies would also need to provide a detailed summary of the datasets used for training. CRCAT would divide collected royalties proportionally across different classes of copyrighted works based on how extensively each category was used. Year Ender 2025: From Tesla’s Entry in India to 'WhatsApp Alternative' Arattai App Launch, Major Tech-Auto Developments.

If disputes arise, the burden of proof would lie with AI developers to demonstrate that they have complied with the royalty requirements. The paper also dismisses voluntary licensing arrangements, including high-profile deals like OpenAI’s agreement with the Associated Press. It argues that such one-to-one partnerships create lengthy negotiations and high transaction costs, ultimately benefiting only major players. Smaller creators and start-ups, it warns, would be left with limited access and little bargaining power under this model.