The Delhi High Court has reserved its order on a plea by news agency Asian News International (ANI) seeking to restrain OpenAI from using its copyrighted content to train ChatGPT, in what is set to become India’s first judicial test of how the Copyright Act applies to artificial intelligence systems.
Justice Amit Bansal concluded hearings after 32 sittings spanning from November 2024, when the suit was filed, to March 2026, with the case evolving into a high-stakes, sector-wide dispute carrying significant implications for media, publishing, and the future of generative AI in India.
Read Full Story
ANI has accused OpenAI of scraping and using its copyrighted news articles without permission or licencing, arguing that ChatGPT not only trained on its content but also reproduced portions of its reports in user responses.
The agency contended that such use falls outside the “fair dealing” exemption under Indian copyright law, particularly as ChatGPT operates as a commercial product.
It further argued that infringement occurs at the very stage of data ingestion, asserting that processes such as scraping, tokenisation, and storage involve unauthorised use of protected expression. ANI also claimed that some of the material appearing in ChatGPT outputs was behind paywalls and accessible only to subscribers, strengthening its case on unauthorised use.
OpenAI Denies Infringement
OpenAI challenged the maintainability of the suit, arguing that Indian courts lack jurisdiction as neither the company nor its servers are based in India. On merits, it maintained that its models are trained on publicly available data, including content hosted on ANI’s website, and that the training process extracts “non-expressive” elements, statistical patterns rather than copyrighted text.
The company also argued that the news agency had failed to demonstrate verbatim reproduction of its content and informed the court that it had blocked ANI’s domain from future training datasets. It maintained that any storage during training is temporary and incidental, falling outside the scope of copyright infringement.
The dispute drew six intervenors, rallying on both sides. The Digital News Publishers Association (DNPA), the Indian Music Industry (IMI), and the Federation of Indian Publishers supported ANI, warned that large-scale unauthorised ingestion of content could undermine the economic foundations of creative industries.
They argued that AI training involves deliberate copying at scale and that outputs generated by such systems could act as substitutes for original works, causing market harm. They also contended that exceptions under Section 52 of the Copyright Act, particularly those relating to digital storage, are narrowly tailored and do not extend to commercial AI systems.
On the other side, the Broadband India Forum, Flux Labs AI Private Limited, and IGAP Project LLP backed OpenAI’s position, advocating for a legal interpretation that accommodates generative AI innovation. They argued that large language models (LLMs) do not reproduce content in a material form and that summarisation or transformation of publicly available data does not amount to infringement. The role of Retrieval-Augmented Generation (RAG) systems was also discussed as potentially non-infringing.
Key Legal Questions Before Court
At the heart of the case lies a fundamental question: whether intermediate copies created during AI training qualify as “transient or incidental storage” under Indian copyright law, or whether they amount to infringement.
ANI and supporting intervenors argued that Indian law does not permit non-infringing intermediate copies beyond narrowly defined exceptions. OpenAI, however, maintained that temporary storage used to extract non-expressive features is legally permissible and essential to machine learning.
The scope of the “fair dealing” exemption under Section 52 also emerged as a major point of contention. OpenAI characterised AI training as a form of private research, while ANI opposed any expansion of the exception to cover commercial AI systems.
Court-appointed amici curiae offered divergent opinions, with one concluding that OpenAI’s actions constituted infringement, while another held that storage and processing for training purposes could fall within permissible limits.
The High Court’s forthcoming decision on interim relief is expected to provide early judicial guidance on how India may regulate the use of copyrighted material in training generative AI systems.
Legal experts believe that the outcome may influence not just this case but also future disputes involving AI developers, content creators, and digital platforms in India.
– Ends
Published On:
Apr 1, 2026 23:18 IST
Tune In
