OpenAI just won the first real courtroom battle in India's AI copyright war. On 24 July 2026, Justice Amit Bansal of the Delhi High Court refused to grant an interim injunction to Asian News International (ANI), India's largest news agency, which had sued OpenAI for allegedly using its copyrighted news reports to train ChatGPT. The court held that OpenAI's temporary storage of copyrighted material during AI training is protected under India's fair-dealing exceptions and does not amount to copyright infringement at this preliminary stage.
The decision is not final. ANI's main copyright suit continues, and the court's observations are explicitly prima facie. But the interim order is already sending shockwaves through India's media industry, the global publishing business, and every AI company that has ever scraped a website without asking. It is also the most significant legal ruling yet to come out of the Global South on the single question that will shape AI's next decade: who owns the data that trains the machines?
What the Court Actually Ruled
ANI sued OpenAI in November 2024, accusing the San Francisco company of copying its news reports without permission to train ChatGPT. The case quickly became a test case for the entire Indian content industry. The Federation of Indian Publishers (FIP), the Digital News Publishers Association (DNPA), and the Indian Music Industry (IMI) all joined as intervenors, arguing that storing and using copyrighted works for AI training without a licence is straightforward infringement. The music industry's side included T-Series, Saregama India, and Sony Music India. The publisher side counted HT Media, The Hindu, India Today, Network18, NDTV, ABP Group, Dainik Jagran, and Times Internet among its members.
Justice Bansal rejected their arguments at the interim stage. He ruled that OpenAI's temporary storage of copyrighted literary works during LLM training is protected as fair dealing under Section 52 of India's Copyright Act. He found that ChatGPT's outputs were not substantially similar to ANI's copyrighted works and that ANI had failed to show that the chatbot memorised or reproduced its original reporting. He also noted that ANI had not demonstrated any actual market harm or loss of subscriptions from OpenAI's use of its content.
Crucially, the court accepted that the balance of convenience favoured OpenAI. Granting an injunction, it said, would harm not only OpenAI but also the public that benefits from AI services in research, education, accessibility, and knowledge dissemination. The court also confirmed that Indian courts have jurisdiction over the dispute even though OpenAI's servers are in the United States, because ChatGPT is accessible in India and the alleged harm was suffered there.
The message was clear: at the interim stage, the public interest in having working AI tools outweighs the publishers' interest in stopping training on their content.
The Policy Fight Behind the Legal One
The Delhi High Court's ruling is not happening in a vacuum. It directly contradicts the policy direction being considered by the Indian government itself. A committee appointed by the Department for Promotion of Industry and Internal Trade (DPIIT) had recommended a compulsory licensing framework under which AI companies could use copyrighted works to train their models by paying licence fees to copyright owners. Ankit Sahni, counsel for the Federation of Indian Publishers, told Mint that the judgment "departs from the policy direction being considered by the government."
This tension — between courts and regulators, between innovation and rights holders, between American tech companies and local content industries — is now playing out across the developing world. India is not the only country grappling with it. Brazil, Indonesia, South Africa, and Nigeria are all watching. Each is trying to decide whether to treat AI training as fair use, fair dealing, or a new activity that requires licensing and payment. The Delhi ruling gives OpenAI and its rivals a powerful precedent in the world's most populous country and one of its fastest-growing digital economies.
For Indian publishers, the ruling is a warning. The court did not say OpenAI is innocent forever. It said ANI had not proven enough at this early stage to shut the service down. But the practical effect is the same: OpenAI can keep training, keep operating, and keep answering questions that may draw on Indian news reporting, while the case drags on for years. Litigation is expensive. News organisations are not. OpenAI can afford to wait.
Why This Matters Beyond India
The global AI copyright debate has been dominated by Western cases. The New York Times sued OpenAI in the United States. Getty Images sued Stability AI in the United Kingdom. Authors and artists have filed class actions in California and London. The European Union's AI Act includes transparency requirements and allows rightsholders to opt out of text-and-data mining for commercial purposes. The general assumption in the industry has been that Western courts and regulators would set the rules, and the rest of the world would follow.
The Delhi High Court ruling complicates that story. India has now staked out a position that is more permissive than the EU's opt-out regime and more favourable to AI companies than the uncertain US fair-use doctrine. If the ruling holds, India becomes a jurisdiction where training on copyrighted material is presumptively legal as long as the model does not substantially reproduce the source. That is a magnet for AI labs looking for legal certainty and a warning to publishers who assumed the world would eventually converge on licensing.
It also fits a broader pattern of AI nationalism. Every major country is now deciding whether to prioritise building domestic AI capability or protecting domestic content creators. China has heavily regulated its own AI sector while encouraging state-backed models. The European Union has built a rights-heavy framework that slows down model development. The United States has so far relied on industry-friendly interpretations of fair use. India's ruling suggests it is choosing the same path as the United States: let AI companies train first, litigate later, and keep the innovation engine running.
For the Global South, this is a high-stakes choice. Countries like India need AI for education, healthcare, agriculture, and governance. They also need local media industries to survive. The Delhi court has implicitly decided that the first need is more urgent than the second, at least for now.
What Happens Next
ANI has already said it will review the judgment and consider its legal options, including potential appeals. Ronil Goger, managing partner at Blaze Legal, told Mint that ANI can appeal the interim order before a division bench of the Delhi High Court while the main suit continues. The main suit will eventually require technical evidence on how exactly ChatGPT was trained and whether OpenAI's models memorised or reproduced ANI's content. Given the novel and far-reaching questions involved, the dispute may ultimately reach the Supreme Court of India.
In the meantime, lawyers familiar with the case expect publishers and music companies to file separate copyright infringement suits against OpenAI in the coming months. They believe they may have stronger claims based on the specific use of their content. The Delhi ruling also does not stop the Indian government from passing a compulsory licensing law if it chooses to override the courts. But the political economy of that choice has just become harder. The court has now framed AI training as a public-interest activity, and any law that imposes heavy licensing costs will face that argument head-on.
OpenAI, for its part, has not publicly commented on the ruling. The company has consistently denied that its models reproduce copyrighted works at scale and has argued that its training on publicly available data is legally protected. In India, the court has accepted that argument for now.
🔥 Hot Takes
1. India just chose cheap AI over local journalism — and the Global South is about to make the same bargain. The Delhi court's public-interest argument sounds noble, but the practical effect is to let a foreign tech giant train on Indian journalism without paying for it. Every developing country now faces the same temptation: free AI now, dying media later. The history of the internet suggests they will take the deal and regret it in twenty years.
2. This is a stealth WTO moment: India is signalling it will not let Western copyright rules block domestic AI development. The ruling is as much about industrial policy as it is about copyright. India wants AI models that understand Indian languages, Indian law, and Indian news. If that means loosening copyright enforcement for training data, so be it. Delhi is effectively telling Silicon Valley: you can build here, but don't expect us to become a copyright fortress for the West.
3. Publishers will lose this war unless they build their own models — licensing fees are a retirement plan, not a strategy. ANI and its allies are fighting the last war. Asking OpenAI for licence fees is fine, but it does not solve the structural problem: AI models will soon generate news-like content that competes with the agencies themselves. The only real defence is to build or partner on Indian-language models trained on proprietary archives. If publishers do not become AI players, they will become AI inputs.
The Bottom Line
The Delhi High Court has handed OpenAI a tactical victory, but it has not ended the war. India's first major AI copyright case will continue through appeals, technical hearings, and possibly the Supreme Court. Governments around the world are still writing their own rules. Publishers are still figuring out how to survive in a world where machines can summarise, rewrite, and replace their reporting.
What the ruling does change is the momentum. For now, in the world's largest democracy, training AI on copyrighted news is fair dealing. The implication is global: as AI nationalism spreads, countries will choose sides not just on chips and data centres, but on who owns the knowledge that feeds the machines. India has made its choice. The rest of the world is watching.