India Proposes Mandatory Royalty System for AI Training on Copyrighted Works

Background and Proposal

India’s Department for Promotion of Industry and Internal Trade (DPIIT) issued a draft framework that would establish a mandatory royalty system for artificial‑intelligence companies that train large language models on copyrighted content. Under the plan, AI firms would gain automatic access to all lawfully available copyrighted works but would be required to deposit royalties into a newly created collecting body. That body would act as a “single window” for rights‑holding organizations and would distribute the proceeds to both registered and unregistered creators.

Rationale

The committee behind the proposal argues that a blanket license would lower transaction costs for AI developers and provide a clear, predictable path for compensating creators. By avoiding a protracted legal battle over whether training data constitutes fair use, the system is presented as the “least burdensome way to manage large‑scale AI training.” The framework also reflects the growing importance of India as a market for generative‑AI tools, noting that major AI firms derive significant revenue from Indian users while relying on Indian creators’ work to train their models.

Industry Response

Technology industry bodies, including Nasscom and the Business Software Alliance (BSA), have submitted formal dissenting comments. Nasscom recommends a broad text‑and‑data‑mining (TDM) exception that would allow AI developers to train on copyrighted material as long as the source is lawfully accessed, arguing that a mandatory licensing regime could slow innovation. The BSA echoes this concern, warning that limiting training data to only licensed or public‑domain works could reduce model quality and increase bias. Both groups suggest that creators who object should be able to opt out rather than being forced into a blanket payment system.

Legal Context

The proposal arrives amid global legal disputes over AI training practices. In India, a lawsuit by news agency ANI against OpenAI has prompted the Delhi High Court to examine whether AI training constitutes a reproducing act or falls under “fair dealing.” Similar controversies are unfolding in the United States and Europe, where courts and regulators are weighing the applicability of fair‑use doctrines to AI model training.

Next Steps

The Indian government has opened the draft for public consultation, giving stakeholders a 30‑day window to submit comments. After reviewing the feedback, the committee will finalize its recommendations, which will then be considered by the broader government. OpenAI and Google have not provided comments on the proposal.