India Mandates Royalties for AI Companies Using Copyrighted Content

The Indian government has proposed a landmark framework for AI training, recommending that all AI companies pay royalties to creators for the use of copyrighted material. The royalty rates would be set by a government-appointed committee and collected through a single collective body of copyright holders.

This move is aimed at ensuring fair compensation to musicians, authors, and news publishers whose work is used for AI model training.

The framework, detailed in a paper titled ‘One Nation, One License, One Payment: Balancing AI Innovation and Copyright,’ was released by a committee under the Department for Promotion of Industry and Internal Trade (DPIIT).

While currently a working paper open for stakeholder feedback, if implemented, it would make India the only country to prescribe a statutory licensing regime for AI developers with government-determined royalties.

Blanket Licensing Over Voluntary Deals

Why Voluntary Licensing Was Rejected

The committee rejected voluntary licensing agreements—such as OpenAI’s licensing deal with the Associated Press—citing concerns over high transaction costs, long negotiations, and unequal bargaining power that could disadvantage small creators and startups. Instead, a mandatory blanket license ensures broad access to copyrighted works while providing equitable compensation.

“The report recommends AI companies have non-discriminatory access to copyrighted work, and takes care of rate fixation with an assurance that all copyright holders are compensated in an equitable way. We have made sure that not a few big companies corner royalties,” said a committee member speaking on condition of anonymity.

The Hybrid Model of Royalty Payment

The DPIIT-led committee recommends a hybrid model under which a mandatory blanket license ensures that all lawfully accessed copyrighted works are paid for, with remuneration rights for the creators and copyright holders.

Legal Access is Mandatory

AI developers cannot use the blanket license to bypass paywalls or circumvent technological protections. Lawful access to content is a prerequisite, ensuring that existing copyright protections remain intact.

Formation of CRCAT

A Copyright Royalties Collective for AI Training (CRCAT) will be established to facilitate royalty collection. CRCAT will:

• Be formed by rightsholders and approved by the Central Government.

• Serve as a centralized body for royalty management.

• Have organizational membership only, with one member per class of work.

• Expand over time to include unorganized sectors without registered copyright societies.

Determining Royalty Rates

Government-Appointed Committee

Royalty rates will be determined by a panel consisting of senior government officers, legal experts, economists, technical specialists, CRCAT members, and AI developer representatives. The rates will:

• Be based on market data, stakeholder input, and economic analysis.

• Be fair, predictable, and transparent.

• Be reviewed and adjusted every three years to reflect technological and market changes.

Flat Rate and Revenue-Based Models

The report recommends flat rates, potentially calculated as a percentage of gross global revenue (excluding taxes) earned by AI developers from systems trained on copyrighted content. These rates may also be subject to judicial review.

“Many AI Developers have built AI Systems which are commercially successful and generating huge revenues. In order to ensure fairness and accountability, such AI Developers must be required to pay royalties to copyright owners for past usage of their works. This is not a punitive measure, but a corrective mechanism to help create a balance in the creative ecosystem,” the report states.

Retrospective Application

AI developers who have already used copyrighted works in training and are monetizing those systems will be required to pay royalties retroactively. This ensures fairness in the creative ecosystem and holds revenue-generating AI developers accountable for past usage.

Division of Royalties

AI companies must submit a ‘Sufficiently Detailed Summary’ of their training datasets to CRCAT, including:

• Type of data (text, images, music, etc.)

• Source of data (online, proprietary, public datasets, publications, etc.)

• Nature of data usage

CRCAT will apportion royalties proportionally among copyright holders based on usage in AI training, ensuring that heavily used categories like music, audiovisual content, or literary works receive appropriate compensation.

In disputes, AI developers will have the initial burden of proof to show compliance. The law presumes the copyright holder’s claim is valid unless proven otherwise.

Conclusion

India’s proposed AI royalty framework marks a first-of-its-kind statutory system globally, ensuring creators are fairly compensated for AI training use. By establishing CRCAT and a government-set royalty panel, the initiative aims to balance innovation and copyright protection. While some in the industry raise concerns over fixed rates, the model promises transparency, equity, and accountability for AI companies, creators, and copyright holders alike.