The Department for Promotion of Industry and Internal Trade (DPIIT) has released a working paper on GenAI training that brings together written submissions from broadcasters, news publishers, music labels and film producers based on separate consultations held with stakeholders of the content industry. The paper, titled ‘One National One License One Payment – Balancing AI Innovation and Copyright’, aims to establish a transparent and unified licensing system for AI training on copyrighted materials while ensuring that content rights holders are paid fair royalties.
Industry groups told a DPIIT committee that AI developers are using copyrighted material at scale without consent, undermining monetisation and eroding competitive advantages. “The dispute is about whether AI can be permitted to scrape content and use it to train an LLM (large language model),” said Abhishek Malhotra, senior advocate at theSupreme Court.
Platform for Direct Permissions
“The content then generates answers to user queriesThis issue is pending before courts across the world, including in India,” Malhotra said. He said a fair, reasonable and non-discriminatory (FRAND) licensing regime can do away with litigations and disputes. “However, this will require dialogue and discussion among various stakeholders, with the government mediating, to arrive at a resolution,” Malhotra said.
The News Broadcasters and Digital Association, which represents television and digital news networks, said AI systems that scrape and repurpose news in real time directly compete with publishers. The group warned that since news content has a narrow monetisation window, summarisation tools that bypass the original source cause what it called irreparable financial harm. The Indian Broadcasting and Digital Foundation, representing broadcasters and digital and OTT platforms, called for the creation of licensing platforms so that AI developers can obtain direct permissions from copyright owners.
It said the law should require voluntary, opt-in licences based on explicit consent, proper attribution and mutually agreed compensation for any use of copyrighted material for training or deployment. News agency ANI noted that global AI developers are already entering commercial licensing agreements with international publishers such as the Financial Times and the Associated Press. ANI argued that Indian content creators must also benefit from these deals. The agency has filed India’s first major AI copyright lawsuit against OpenAI.Data Mining Exceptions
The Indian Music Industry asked the government to retain the country’s current copyright regime, which supports free market licensing, and warned against introducing text and data mining exceptions that could weaken protections.
The DPIIT committee noted that a fully voluntary licensing system may not be feasible at internet scale as online data is projected to reach 175 zettabytes by 2025, potentially involving content belonging to a billion separate copyright owners. Part 1 of the working paper reviews global options such as blanket exemptions, text and data mining exceptions, opt-out mechanisms, voluntary licensing, and collective licensing. The panel concluded that none of these approaches fully addresses India’s needs. It rejected a zero price licence for AI training, saying it would weaken incentives for human creativity and reduce the production of original content.
Instead, the paper proposes a hybrid system. Under this model, AI developers would automatically receive a blanket licence to train on all lawfully accessed content, without needing individual negotiations. They would pay royalties only when the AI tool is commercialised. A government-appointed committee would set royalty rates, which may be reviewed by courts. A centralised body would collect and distribute the royalties to reduce costs and ensure clarity for both large and small developers, it said. Legal experts said as multiple AI tools emerge, it will be increasingly difficult for any publisher, broadcaster or music label to identify and stop the unauthorised use of their content since AI training occurs globally across jurisdictions.
As AI evolves, the training use of copyrighted works should also be recognised as a form of adaptation to safeguard creators’ rights, said Mallika Noorani, senior partner at Parinam Law Associates. “However, meaningful protection will ultimately depend on enforceability, which requires far greater auditability, transparency and cooperation from AI developers,” she said. Kaushik Moitra, partner and practice lead for regulatory, IP and TMT at Bharucha and Partners, said broadcasters and other content creators can pursue infringement actions for unauthorised scraping under the broadcast reproduction right.
