Sign In
By signing in or creating an account, you agree with Associated Broadcasting Company's Terms & Conditions and Privacy Policy.
By signing in or creating an account, you agree with Associated Broadcasting Company's Terms & Conditions and Privacy Policy.
New Delhi: India is trying to answer one of the toughest questions in artificial intelligence today. Who gets paid when AI is trained on human creativity that already exists on the internet. The government has taken a first step toward that answer with a new working paper that looks at AI training, copyright rights and the future of India’s creative economy.
The Department for Promotion of Industry and Internal Trade published the document on Monday and is now asking the public to respond within 30 days. This is not the final set of rules, but it shows where policy thinking inside the government is heading. And that direction could affect every AI company building large language models in India, from big tech to student-led startups.
The core of the proposal is simple on paper. AI developers would get mandatory access to all copyright protected content that they can lawfully reach online. They would not need to ask each creator for permission. But they must pay royalties when the model is commercialised. Those payments would go to a single non profit collecting body run by rights holders and designated by the government.
A committee formed in April examined policies across the US, EU, UK and Singapore before arriving at this idea. It argues that India needs a special system that both supports AI growth and protects creative livelihoods. The paper says that long negotiations and legal uncertainty can slow down young AI companies, especially startups working under India’s IndiaAI Mission.
The proposal rejects what the technology industry calls a text and data mining exception. This exception would let AI models scrape publicly available content for free. The committee says that would “undermine copyright” and make smaller creators powerless to seek payment. It also dismisses opt out models where creators can forbid the use of their content. According to the paper, most small creators lack tools to enforce an opt out. Even if they did, scraped content often loses metadata making it hard to track later reuse.
According to The Hindu, Nasscom disagrees with the government side. It calls the royalty system a “tax on innovation” and suggests allowing mining for publicly accessible content without paywalls with an option for creators to reserve their work. Nasscom’s view is that rights holders need protection but large scale licensing could hurt the speed of innovation.
The committee instead focuses on ensuring:
• fair compensation for copyright holders
• reduced compliance cost for AI startups
• no need for contract by contract negotiations
• a level playing field between big and small developers
Royalties would be decided by a government appointed committee and could be challenged in court if needed.
The working paper lands at a time when Indian companies are aggressively building their own AI systems. Krutrim AI, started by Bhavish Aggarwal, is training a homegrown foundational model. Sarvam AI is working on culturally rooted systems like OpenHathi. Government backed teams are developing multilingual models through IIT Bombay and BharatGen Technology Foundation. Tech Mahindra is working on an Indic model called Project Indus.
All these models need huge volumes of data. And that is where this policy could make their lives easier by removing legal risk linked to scraping data used to train LLMs.
Delhi High Court is already hearing a case filed by the Digital News Publishers Association against OpenAI, accusing it of using copyrighted news articles to train ChatGPT. Globally, The New York Times is in a fight with OpenAI and Microsoft, and writers and musicians have dragged AI firms to court in the US and UK. The DPIIT paper says India cannot wait for these cases to end.
By creating “one nation, one licence, one payment,” the government says it can protect creators while supporting AI development.
