Introduction:
In ANI Media Pvt. Ltd. v. OpenAI Inc. & Another, the Delhi High Court delivered a significant interim ruling in India’s first copyright dispute involving ChatGPT and generative artificial intelligence. Justice Amit Bansal dismissed the interim injunction application filed by ANI Media Pvt. Ltd., holding, on a prima facie assessment, that OpenAI’s storage of ANI’s copyrighted works for training its artificial intelligence models does not amount to copyright infringement under the Copyright Act, 1957. The Court further held that the responses generated by ChatGPT were not substantially similar to ANI’s original works and, therefore, did not constitute infringement under Section 51 of the Act.
The dispute arose after ANI alleged that OpenAI had unlawfully used its original news reports to train ChatGPT and that the chatbot occasionally reproduced ANI’s content verbatim. ANI also claimed that ChatGPT generated false statements attributed to the news agency, commonly referred to as “hallucinations,” thereby harming its credibility and creating the risk of misinformation. The suit sought an interim injunction restraining OpenAI from using ANI’s copyrighted material during the pendency of the proceedings.
The case assumes considerable importance because it is the first litigation in India examining the relationship between copyright law and generative AI technology. It raises fundamental questions regarding the scope of copyright protection, the doctrine of fair dealing, the legality of AI model training and the liability of AI systems for inaccurate or fabricated outputs.
Arguments of the Parties:
ANI contended that it owns copyright in its original news reports, articles and other journalistic content. It alleged that OpenAI had copied and stored these works without authorisation to train its large language models, thereby commercially exploiting its copyrighted material without obtaining any licence or permission.
According to ANI, ChatGPT was capable of reproducing portions of its news reports in response to user queries, amounting to unauthorised reproduction of protected literary works. The news agency argued that such use directly infringed its exclusive rights under the Copyright Act, particularly because OpenAI commercially benefited from the use of copyrighted content while developing and operating its AI systems.
ANI further alleged that ChatGPT occasionally generated incorrect information while attributing it to ANI. These “hallucinations,” according to the plaintiff, seriously damaged the reputation of a professional news organisation by creating false reports in its name. It argued that the circulation of inaccurate information attributed to ANI could mislead readers, erode public confidence in journalism and contribute to the spread of fake news.
Seeking urgent interim protection, ANI requested the Court to restrain OpenAI from storing, processing or using its copyrighted material until the final adjudication of the suit. It maintained that continued use of its content would cause irreparable injury that could not be adequately compensated by damages alone.
OpenAI opposed the grant of any interim injunction. At the threshold, it questioned the maintainability of the suit before Indian courts by arguing that its AI models are developed and trained outside India and that the relevant servers are located in the United States.
On merits, OpenAI submitted that the machine learning process is fundamentally transformative and does not merely reproduce copyrighted works. It argued that AI training involves statistical analysis of large datasets to identify linguistic patterns rather than copying expressive content for commercial republication.
OpenAI also contended that copyright protects the original expression of ideas but not the underlying facts contained in news reports. Since news primarily communicates facts and events, the factual information itself cannot be monopolised through copyright protection.
The company further pointed out that publishers unwilling to have their content used for AI training may request inclusion in a blocklist to prevent future crawling of their websites. It argued that adequate technological mechanisms already exist for content owners who wish to restrict access to their publicly available material.
Court’s Judgment:
Justice Amit Bansal declined to grant the interim injunction sought by ANI and held that the plaintiff had failed to establish a prima facie case warranting immediate restraint against OpenAI. While clarifying that the observations were only tentative and would not affect the final adjudication of the suit, the Court expressed the prima facie view that OpenAI’s conduct did not presently amount to copyright infringement under the Copyright Act.
The Court first considered whether the storage of ANI’s copyrighted material during the AI training process constituted infringement. It observed, prima facie, that such storage falls within the scope of Section 52(1)(a) of the Copyright Act, 1957, which recognises the doctrine of fair dealing. Under this provision, fair dealing for purposes such as private or personal use, research, criticism, review and reporting of current events does not amount to copyright infringement.
On a prima facie analysis, the Court held that OpenAI’s storage of copyrighted works for training its AI models attracted the protection available under Section 52(1)(a). Consequently, such storage could not presently be regarded as an infringing act under Section 51 of the Copyright Act.
The Court also examined ANI’s allegation that ChatGPT reproduced its original reports. It observed that the outputs generated by ChatGPT were not substantially similar to ANI’s copyrighted works. Since copyright infringement requires unlawful reproduction of a substantial part of the original expression, the absence of substantial similarity weighed against the grant of interim relief.
Justice Bansal therefore concluded, at the interim stage, that the responses generated by ChatGPT did not amount to infringement under Section 51 of the Copyright Act. The Court was not satisfied that the plaintiff had demonstrated a prima facie violation of its exclusive copyright.
The Court’s observations indicate an important distinction between the process of training an AI model and the reproduction of copyrighted material. Merely storing or processing data for machine learning, according to the Court’s prima facie assessment, does not automatically constitute infringement where the statutory defence of fair dealing is attracted.
Although ANI also complained about AI-generated “hallucinations” attributing false statements to the news agency, the interim order primarily addressed the copyright issues raised under Sections 51 and 52 of the Copyright Act. The detailed reasoning regarding other aspects of the dispute will become clearer once the complete judgment is released.
By refusing interim relief, the Court has allowed the main suit to proceed to trial, where complex questions relating to copyright protection, AI training, generative technologies, fair dealing and digital innovation will be examined in greater detail. The final decision is expected to become a landmark precedent governing the interaction between artificial intelligence and copyright law in India.
The interim ruling reflects the judiciary’s cautious approach in balancing the rights of copyright owners with technological innovation. While recognising the legitimate concerns of content creators, the Court declined to impose immediate restrictions on AI development in the absence of a clear prima facie case of copyright infringement. The litigation is likely to shape the future legal framework governing generative AI, digital content and intellectual property rights in India.