A Delhi court refuses an injunction against OpenAI
On 24 July 2026 the High Court of Delhi dismissed ANI Media's application for an interim injunction against training ChatGPT on its content: Justice Amit Bansal held, on a prima facie basis, that OpenAI's storage of ANI's works for training the LLMs behind ChatGPT falls under fair dealing, Section 52(1)(a) of India's Copyright Act 1957, and so is not infringement under Section 51, and that ChatGPT's RAG-generated responses are not substantially similar to ANI's original text.
Why it matters
A high court ruled for the first time in India on whether training a language model on copyrighted text can fall under the fair-dealing provision, and found that it can — provisionally, within one interlocutory application, not as a final ruling on the suit. The same month, a Munich court in GEMA v. Suno reached the opposite conclusion about training on music (evt-0397): one month produced two courts reading training on protected material in opposite ways.
The case is ANI Media Pvt. Ltd. v. Open AI OpCo LLC, CS(COMM) 1028/2024, I.A. 45300/2024, from a suit filed in November 2024. The order was pronounced on 24 July 2026 (judgment was reserved on 27 March 2026), electronically signed at 17:17:24. The court held it has territorial jurisdiction (paras 43-48): ANI's registered office is within the court's jurisdiction, OpenAI specifically targets and offers services to users in India, and RAG-generated responses are produced within the jurisdiction — sufficient under Section 62(2) of the Copyright Act and Section 20 of the Code of Civil Procedure. The court denied ANI's application for an interim injunction (I.A. 45300/2024) on these grounds (paras 262-273): ANI could have blocked its website to third-party crawlers, including OpenAI, but had not done so, while OpenAI stated it had already blocked ANI's site from both training and RAG search; ANI had itself offered OpenAI a licence to all its content for USD 7.5 million by a letter of 3 October 2024, making its claim quantifiable and compensable in damages, unlike OpenAI's loss if enjoined; an injunction would affect millions of ChatGPT users in India, most of them unpaid, a public-interest factor courts must weigh in intellectual-property cases. The court's prima facie finding (para 271): OpenAI's storage of ANI's original literary works to train the LLMs underlying ChatGPT falls under Section 52(1)(a) of the Copyright Act and so is not infringement under Section 51; outputs generated through RAG are not substantially similar to ANI's original works; ANI failed to establish memorisation or regurgitation of its text by the model. What the record does not claim. That the case is decided: paragraph 274 states the observations bear only on this application, not on the final outcome of the still-pending suit. That the ruling extends beyond India or binds other courts: this is a first-instance interlocutory order, not an appellate or final judgment. Every part of the 135-page order, including its fuller treatment of memorisation beyond the finding summarised here: this record does not restate it in full.