Imagine teaching a computer to paint unique masterpieces, write gripping novels, or compose music that touches the soul. To make that possible, the machine must be “fed” millions of pages of text, thousands of hours of audio recordings, and gigabytes of images. It must quite literally absorb human culture.
That is exactly how generative artificial intelligence (AI) works. And this is where we run straight into one of the most controversial legal questions of our time: do technology companies have the right to train their AI models on the works of real creators free of charge and without permission?
Until recently, this battle seemed to be unfolding somewhere far away—in courtrooms across the United States and Europe. But the summer of 2026 changed everything. India has found itself at the center of its own legal thriller: the Delhi High Court is currently hearing the case of ANI Media v. OpenAI. It is the country’s first major precedent that could reshape the rules of the game worldwide.
A Global Cheat Sheet: How Other Countries Are Addressing the Issue
While India is searching for its own path, the rest of the world has already taken positions:
- United States (“fair use” doctrine): Tech giants argue that AI does not steal—it transforms knowledge. They claim that an AI model learns much like a person who reads a hundred books before writing one of their own. Artists, however, strongly disagree, calling it piracy on a cosmic scale.
- European Union and the United Kingdom (“opt-out” model): Here, the rule is different. AI companies may collect data from the internet unless the author has explicitly and technically opted out by indicating that their works may not be used.
India has taken a different approach. Its Copyright Act of 1957 is considerably stricter. Unlike U.S. law, it contains no flexible “fair use” doctrine. Instead, it provides a narrow and exhaustive list of exceptions. The large-scale downloading of articles, books, or songs to train commercial AI systems clearly falls outside those exceptions.
Courtroom Drama: A News Agency vs. ChatGPT
What is the lawsuit about?
News giant ANI accuses OpenAI, the creator of ChatGPT, of secretly copying its massive archive of journalistic content to train large language models.
OpenAI’s defense follows a familiar line: “We only used publicly available information, we learn facts rather than authors’ creative expression, and in any event, we have already blocked AI access to your website.”
However, the proceedings have exposed three fundamental problems where twentieth-century laws struggle to cope with twenty-first-century technologies:
- Temporary Memory or Copyright Infringement?
Tech companies argue that data is copied only “momentarily” during processing. Publishers counter that storing gigabytes of copyrighted material throughout the training process constitutes continuous and unlawful reproduction. - The “Parrot” Effect
OpenAI maintains that AI merely identifies statistical relationships between words. ANI, however, presented examples where ChatGPT reproduced the agency’s paywalled articles almost word for word, depriving publishers of revenue. - AI Hallucinations
AI systems sometimes fabricate news stories and falsely attribute them to reputable media organizations, damaging both their reputation and their brands.
The Government’s Proposal: “Compulsory Licensing” as a Compromise?
Since even court-appointed experts have been unable to reach a consensus, India’s Department for Promotion of Industry and Internal Trade (DPIIT) has proposed a third option—a system of compulsory licensing.
The idea is straightforward: AI developers would be allowed to train their models on virtually any dataset without first negotiating individual permissions from every rights holder. In return, companies would be required to pay government-regulated royalties into a dedicated compensation fund for creators.
The proposal immediately divided public opinion into two opposing camps:
| Tech Optimists | Copyright Advocates |
|---|---|
| Innovation depends on data. Without access to local languages and cultural content, Indian startups will never develop competitive domestic AI systems and will inevitably lose to Silicon Valley. | This undermines freedom and property rights. Ownership includes the right to say “no.” Creators should not be forced to surrender their works to technologies that may eventually replace them. |
India’s Hybrid Approach
India is unlikely to simply copy either the American or the European model. An overly restrictive approach would cripple local AI startups, while unrestricted freedom for AI developers could severely undermine the creative industries.
Many analysts therefore support a balanced solution: full transparency—requiring AI companies to clearly disclose the data used to train their models—combined with an efficient marketplace for voluntary commercial licensing.
The Delhi High Court’s decision in ANI Media v. OpenAI will reveal which direction the pendulum will swing: toward unfettered technological capitalism or toward a new legal framework designed to protect human creativity from digital absorption.
