What kind of AI economy does India want to build?

Artificial intelligence (rechristened by Trump as Super Intelligence) is evolving so rapidly that not a day passes without at least half dozen news and views on its varied dimensions appearing in the print medium. One should not be surprised that there will be a score of surprises in the future.

No wonder, India’s debate on AI governance is transitioning from ambition to hard choices. At stake is not simply how India regulates AI, but the kind of AI economy that regulation will help shape. The Delhi High Court’s recent judgment in ANI v. OpenAI brings that question into focus. In July, the court took a prima facie view that OpenAI’s training of its Large Language Model on ANI’s copyrighted news material could fall within the fair dealing exception under the Indian Copyright Act. In arriving at this conclusion, the court reached beyond copyright, weighing innovation, access to information, dissemination of knowledge and public interest.

Since then, the judgment has sparked considerable debate among the stakeholders, adding India to the growing list of jurisdictions currently grappling with the regulation of LLM training and its legal consequences.

Also read | India needs AI that no one else can switch off: Andrew Ng

However, as India’s AI architecture begins to take shape, its central policy task should be to decide the economic and innovation outcomes it wants that architecture to produce, to the best extent possible.

India is already investing in building its domestic AI capability. The IndiaAI Mission is focused on reducing the cost of compute, expanding access to data, financing local startups and improving the domestic research infrastructure. Earlier in the year, the Government of India constituted the AI Governance and Economic Group, as a high-level, inter-ministerial institutional mechanism to coordinate Artificial Intelligence policy, governance, and economic strategy in India. This ambition now needs to be supplemented with a legal framework that reconciles the competing regulatory objectives shaping the same ecosystem.

This is unlikely to be achieved through a single statute but through multiple areas of law operating in tandem with each other. Copyright law will settle questions on access to protected training material. Competition law will influence market entry and market power. Data protection law will constrain the use of personal data for training datasets while liability rules will determine the risks borne by developers when models produce harmful outputs. Each area of law addresses a distinct concern, but their effects converge on the same stakeholders and the same market.

Also read | Why is sovereign AI critical to India’s next growth leap?

The tensions surrounding copyright and AI training make this convergence particularly visible. Authors, journalists, publishers and other creators need viable economic incentives to continue contributing to the knowledge economy. If AI developers are allowed to train their commercial models on copyrighted material without the requisite permissions or compensation to the creators, it may financially weaken the creative industries that produce this material.

On the other hand, a stringent licensing regime comes with its own set of economic costs and challenges. The training of sophisticated models requires access to large quantities of data. Licensing at this scale involves multiple considerations such as identifying the rightsholders, negotiating permissions with numerous creators and publishers, paying them royalties and monitoring compliance at various levels. Global AI firms enter these markets with deep pockets and established licensing relationships that domestic startups may not be in a position to compete with.

This may create a regulatory paradox: stronger creator protection could make AI markets less competitive by raising barriers to entry for local companies and entrenching the position of dominant AI companies.

Seen in this light, the policy choice between permitting an expansive use of copyrighted materials and mandating licensing for AI training has implications well beyond copyright law. India should, therefore, assess AI-related legal rules not only by whether they resolve the immediate legal issue, but also by the market structure and economic incentives they are likely to create. Furthermore, it will also need to provide built in flexibility to cover unknown issues which would crop up as the whole thing evolves.

The United Kingdom offers a useful lesson in this regard. In March 2026, its government published a formal Copyright and Artificial Intelligence Impact Assessment report that assessed how the different copyright approaches to AI training would affect the rightsholders, AI developers, smaller firms and the economy at large. Crucially, the government made copyright reform contingent on the chosen approach meeting its broader objectives for the economy and its citizens.

The point is not Britain’s eventual answer, but its effort to define the outcome before settling the instrument.

India does not need to adopt this exact model. It should, however, borrow the method. First, it should define the economic outcomes it wants its AI framework to serve, including sustaining creator incentives, preserving competitive markets, strengthening domestic AI capabilities and advancing the wider public interest. It should then compare the likely effects of different legal approaches against those objectives before deciding what the law should require.

The same outcome-first approach should extend beyond copyright. Competition interventions, data-protection rules and safety obligations should also be assessed for their cumulative effects on entry, investment, innovation and public welfare. Regulators may have different mandates, but AI-related rules should be assessed against a common set of economic objectives so that their combined effects support, rather than undermine, India’s wider AI ambitions.

The AI Governance and Economic Group gives India a practical institutional anchor for this exercise. It could develop a common economic-impact framework as a mandatory checkpoint for significant AI proposals, testing their effects on creator incentives, competition, domestic capability and public welfare before they are adopted.

ANI v. OpenAI is an early sign of what lies ahead: existing laws will increasingly be asked to resolve questions created by technologies they were never designed to govern. Judges will continue to decide individual disputes while the regulators will shape the rules within their own mandates. India’s larger responsibility is to ensure that these decisions align with a coherent economic strategy for AI governance. The shape of India's AI economy should be a deliberate choice, not one it discovers by accident long after the rules have already built it.

Mehta is Secretary General of CUTS International. Gupta is Asstt Professor, Jindal Global University.