India has proposed a mandatory royalty regime for AI companies that train models based on copyrighted content. It’s a move that could reshape the way OpenAI and Google operate in what has become one of the world’s most important and fastest-growing markets.
India’s Ministry of Industry and Internal Trade Promotion on Tuesday announced a proposed framework that would give AI companies access to all copyrighted works for training, with payments distributed to creators, in exchange for paying royalties to a new collection body made up of rights holders. The proposal argues that this “blanket license mandate” would ensure that writers, musicians, artists, and other rights holders are compensated when their works are scraped to train commercial models, while reducing compliance costs for AI companies.
India’s proposal comes amid growing concerns in global markets about how AI companies use copyrighted material to train their models, a practice that has sparked lawsuits from authors, news organizations, artists and other rights holders in the United States and Europe. Courts and regulators are still considering whether such training constitutes fair use, allowing AI companies to operate under legal uncertainty and expand rapidly without clear regulation.
Unlike the United States and the European Union, where policymakers are debating transparency obligations and the boundaries of fair use, India is proposing one of the most interventionist approaches yet: giving AI companies automatic access to copyrighted material in exchange for mandatory payments.
An eight-member committee set up by the Indian government in late April argues that the system avoids years of legal uncertainty while ensuring creators are compensated from the start.
Defending the system, the commission said in a 125-page submission (PDF) that blanket licenses are the least burdensome way to manage large-scale AI training, saying they are “aimed at providing AI developers with easy access to content, reducing transaction costs, (and) ensuring fair compensation for rights holders.” The application adds that a single collection body would act as a “single point of contact”, eliminating the need for separate negotiations and allowing copyright fees to flow to both registered and unregistered creators.
The committee also noted India’s growing importance as a market for GenAI tools. Citing OpenAI CEO Sam Altman’s statement that India is the company’s second-largest market after the United States and “could possibly become its largest,” it argues that while AI companies make a lot of money from Indian users, they rely on the work of Indian creators to train their models, so some of that value should go back to those creators. That is part of the rationale for establishing a “balanced framework” to guarantee compensation.
tech crunch event
san francisco
|
October 13-15, 2026
India’s proposal comes as legal battles rage around the world over whether AI companies can legally use copyrighted material to train models.
In India, news agency ANI filed a lawsuit in the Delhi High Court alleging that OpenAI’s article was used without permission. This case prompted courts to consider whether AI training itself constitutes copying, or whether it is protected by “fair dealing.” Similar disputes are facing U.S. and European courts, with creators claiming the tech companies built their models on unlicensed content.
There is backlash and opposition to AI proposals.
However, not everyone is convinced by the model proposed by the Indian government.
Nasscom, an industry group representing technology companies such as Google and Microsoft, filed a formal objection, arguing that India should instead adopt a broad text and data mining exception that would allow AI developers to train with copyrighted content, as long as the material is accessed legally. Warning that compulsory licensing could slow innovation, opposing rights holders said companies should be allowed to opt out rather than forced to pay for all training data.
The Business Software Alliance, which represents global technology companies such as Adobe, Amazon Web Services and Microsoft, has pressed the Indian government to avoid a purely license-based regime. The report called on India to introduce explicit text and data mining exceptions, arguing that “relying solely on direct or legal licensing of AI training data is impractical and may not yield the best results.”
Limiting AI models to a smaller set of licensed or public domain material can reduce model quality and “increase the risk that the output simply reflects the trends and biases of the limited training dataset,” the BSA warned, adding that a clear TDM exception would better balance innovation and rights holders’ interests.
The commission did not consider both broad text and data mining exceptions and opt-out models, arguing that such systems would undermine copyright protection or be impossible to enforce. Instead, it proposed a “hybrid model” that would give AI companies automatic access to all legally available works, but require them to pay royalties to a central collection body that would distribute revenue to creators.
The Indian government has now launched a public consultation on the proposal, giving companies and other interested parties 30 days to submit comments. After considering the feedback, the committee will finalize its recommendations before the framework is adopted by government.
OpenAI and Google did not respond to requests for comment.
