In a move that could fundamentally rewrite the economic rulebook for artificial intelligence, the Indian government panel has proposed a landmark regulatory framework requiring AI companies to pay mandatory royalties for using Indian content to train their models. The proposal, released Monday by a committee under the Department for Promotion of Industry and Internal Trade (DPIIT), represents a significant setback for tech giants like OpenAI and Google, who have long argued that publicly available internet data constitutes “fair use.”
The panel’s report recommends a “hybrid model” that guarantees AI companies access to data while ensuring creators get paid. If India adopts this legislation, it will have substantial divergences from the United States and Japan and will thus represent an outlier in the carriage of the international digital copyright landscape.
The ‘One Nation, One License’ Model
The main focus of this proposal is a single, streamlined framework to alleviate the operational headaches associated with issuing individual licenses. One option that has been put forward by the Committee is the establishment of a new centralized umbrella organization, which may be referred to as the Copyright Royalties Collective for AI Training (CRCAT).
Under this system, AI developers would be granted a “mandatory blanket license” to use lawfully accessed Indian content. In return for access to copyrighted works, AI companies are required to pay a percentage of their gross revenues to the copyright owners through the Copyright Royalty Collective, which distributes the money. The amount of money due to copyright holders will not be determined by the marketplace, but by a committee appointed by the government based on a percentage of the gross global revenue of the AI company. The intent of this system is to correct a fundamental inequity: many AI businesses profit from the use of billions of data points generated by journalists, authors, artists, etc., while these same individuals earn little or no compensation.
Why ‘Opt-Out’ Was Rejected
The panel rejected the current European Union approach of requiring creators to withdraw their works from training datasets by opting out. In this system, creators would have to take action to request that their works not be included in training datasets for AI.
The Indian committee argued this puts an unfair burden on creators. Tracking down one’s work inside the “black box” of a massive Large Language Model (LLM) is technically difficult and resource-intensive for individual artists or small publishers. The report noted that once data is stripped of its metadata and ingested into a model, control is often “irrecoverable.” The Panel has proposed a universal licensing obligation on the industry so that smaller organisations will not be disadvantaged by private negotiations with multi-billion dollar technology companies.
Retroactive Payments on the Table
The suggestion by the panel that there be retroactive liability for technology companies using past train datasets in their products looks to be the most concerning part of this proposal. If technology companies that trained their AI models on Indian data are found liable for revenue derived from that use, the financial impact on many of them could be significant, especially with India being one of the biggest markets for ChatGPT and Google Gemini. Furthermore, should an AI company create a commercial product from an AI system that was previously trained, the original developers would still need to pay royalties on any revenue generated by the commercial product, regardless of whether or not the training occurred before commercialization.
In effect, this proposal would not only put a significant financial burden on technology companies doing business in India but would also create a disincentive for them to continue to build products from their already-trained AI systems.
Silicon Valley Pushes Back
Leaders of the tech industry have expressed their discontent with this plan due to the immediate backlash created by the proposal. Nasscom, an influential trade organisation and representative of tech behemoths such as Google and Microsoft, has submitted a formal dissent to the committee outlining its objections. It contends that imposing a mandatory fee will serve as a “tax or levy on innovation” and ultimately hinder the startup ecosystem of India, which has great potential for growth.
The Motion Picture Association (MPAA) has been equally vocal in voicing its opposition to the proposed changes to current copyright laws and has instead suggested that new licensing agreements be implemented on a voluntary basis. Senior officials from each of the aforementioned companies in technology fear that if the proposed changes are adopted, they will create an environment within which technology companies will become less likely to elect to develop their AI products in India. In fact, it could also force technology companies to keep Indian data completely separate.
The Legal Backdrop: ANI vs. OpenAI
The government’s aggressive stance mirrors the mood in Indian courtrooms. The proposal comes as OpenAI is already locked in a high-profile legal battle with ANI (Asian News International), one of India’s largest news agencies.
ANI has accused OpenAI of using its copyrighted news feeds to train ChatGPT without permission, and of attributing fabricated quotes to the agency—a phenomenon known as “hallucination.” While OpenAI has defended itself by claiming “fair use” and noting that it blocked ANI’s site after a request, ANI argues that its content remains accessible to the AI via third-party scrapes. The new government proposal would effectively moot OpenAI’s “fair use” defense in India, replacing it with a “pay-to-play” statutory requirement.
