In December 2025, days after Dhurandhar became a box-office hit, the government quietly released a working paper that may end up shaping India’s AI economy for years. The Department for Promotion of Industry and Internal Trade proposed a straightforward bargain: let AI companies train models on copyrighted material they can legally access, but ensure writers, publishers and creators are paid through a central licensing and royalty system.
The proposal showed haunsla (courage), acknowledging a reality policymakers elsewhere still hesitate to confront. Yet, courage is not enough. The plan now needs sustained political will to become a reality. Without follow-through, Indian creators and the wider economy may never see the fair share they deserve from fueling global AI models with their content.
Generative AI systems are being built on vast quantities of human-created content, much of it copyrighted, and the economic value generated from that material cannot remain entirely concentrated in the hands of a few technology companies.
Public consultations were held, and stakeholders weighed in, but the momentum soon faded. India’s publishers, artists and media companies were left waiting, while AI developers continued operating in a regulatory vacuum.
Policymakers fear that hurried legislation could burden startups with compliance costs or leave India tied to rules that may quickly look outdated as the technology evolves. Critics often point to the US, where courts and private licensing deals continue to define the debate in the absence of AI-specific copyright laws. America’s technology dominance, they argue, has hardly suffered from regulatory restraint.
But India is not the US. America can afford ambiguity because its technology firms already command much of the global AI economy. India risks ending up in the worst possible position, where its creative industries are unable to earn from the content powering AI systems, while domestic AI firms continue operating without clear legal guardrails.
That uncertainty is no longer theoretical. In March 2026, the Delhi High Court heard ANI’s case against OpenAI, the country’s first major legal challenge examining whether training AI systems on copyrighted material constitutes infringement. The judgment is still pending, but the broader problem is already evident.
Courts can settle individual disputes, but they are poorly placed to define the rules of an industry changing this rapidly. If judges are left to force-fit generative AI into copyright laws written decades ago, India could end up with conflicting rulings, prolonged litigation and years of uncertainty that serve neither creators nor technology companies particularly well. That outcome helps nobody. Not creators. Not startups. Not investors.
Lost Time
India has already seen how costly policy drift can become. After the Supreme Court’s landmark privacy judgment in 2017, the country spent years debating data protection before finally passing a law in 2023 that arrived far weaker and less precise than initially promised. During that period, businesses operated with little clarity even as other jurisdictions moved ahead with firmer standards. A similar pattern emerged in platform regulation. New Delhi publicly welcomed Australia’s News Media Bargaining Code, which pushed technology companies to compensate publishers, but stopped short of introducing a comparable framework at home. India’s National AI Strategy, unveiled in 2018 with considerable fanfare, eventually narrowed into a set of broad recommendations without meaningful regulatory backing.
Meanwhile, the market moved ahead without India.
OpenAI signed licensing agreements with the Associated Press for access to news archives. Google struck multi-year content agreements with News Corp after Australia’s bargaining framework came into force. These arrangements demonstrate that at least some AI firms recognise the commercial necessity of compensating content owners.
But bilateral deals favour only the largest publishers with the scale and leverage to negotiate. Smaller media organisations, regional publishers and independent creators are left exposed. Without a structured domestic licensing regime, many Indian creators have little bargaining power beyond costly legal action.
The DPIIT paper recognised this imbalance early. Its proposed licensing framework was not anti-technology. If anything, it attempted to prevent a future where endless litigation paralyses innovation. Even many AI developers would privately welcome standardised rules that reduce legal uncertainty and avoid jurisdiction-by-jurisdiction disputes.
The longer India delays, the weaker its negotiating position becomes. AI models are already being trained. Global standards are already emerging. Once entrenched commercial practices and foreign legal precedents solidify, India’s ability to shape outcomes in favour of its creators will narrow sharply.
This is ultimately not just a copyright debate. It is an economic question about who captures value in the AI era. Countries that provide clarity will attract investment, innovation and partnerships. Countries that remain stuck between consultation papers and courtrooms risk watching both capital and leverage drift elsewhere.
India showed rare foresight in recognising the problem early. Now it needs the eendhan (fuel) to act before this opportunity turns into a permanent disadvantage. Policy caution may feel safer in the short term, but prolonged ambiguity carries its own costs. Every month of delay erodes creators’ bargaining power while discouraging responsible AI development.
The real risk is no longer overregulation. It is inertia.
.png)
