In a landmark ruling for AI regulation in India, the Delhi High Court on Friday said that OpenAI’s use of news agency ANI’s content to train its ChatGPT service did not amount to copyright infringement.
News agency ANI Media Pvt Ltd had approached the Delhi High Court in 2024 against OpenAI Inc (OAI) and OpenAI OpCo LLC, accusing the US company of using its published content without permission to train its AI models and of having its chatbot attribute fabricated stories to the news agency.
The court held that OpenAI’s act of storing and using copyrighted works to train Large Language Models (LLMs) underlying ChatGPT does not prima facie constitute copyright infringement, framing the process as falling within the ambit of fair dealing.
In his judgment, Justice Amit Bansal observed that restraining the AI giant from utilising public data for model development would cause irreparable harm not only to the developer but to the wider public interest, the Indian Express reported.
“Irreparable injury would be caused not only to OpenAI but also to the public at large if an interim injunction is granted,” ruled Justice Bansal, according to The Indian Express.
ANI vs OpenAI case: Key highlights of the verdict
Fair dealing and non-expressive use: The court held that storing original literary works to train underlying AI models falls under the fair dealing exception under Section 52 of the Copyright Act, 1957.
No direct output infringement: Justice Bansal noted that ANI failed to demonstrate any systematic “memorisation or regurgitation” of its copyrighted material in ChatGPT’s responses, as the generated outputs were not substantially similar to ANI’s original articles.
Public interest considerations: The bench emphasised the potential chilling effect an interim stay would have on AI technology adoption and public access to emerging technological tools in India.
OpenAI vs ANI case: Why the dispute erupted
The lawsuit, first filed in late 2024 by ANI, marked India’s first major judicial test regarding generative AI and intellectual property. ANI alleged that OpenAI exploited its copyrighted news content without permission or compensation, claiming the AI chatbot reproduced its news stories verbatim and, in some instances, falsely attributed generated statements to the news agency.
OpenAI argued that LLM training operates on structural analysis and statistical patterns rather than raw copying for publication, maintaining that public data processing for machine learning is fundamentally non-infringing.
To navigate the complex legal terrain, the High Court had previously appointed intellectual property advocate Adarsh Ramanujan and law professor Dr Arul George Scaria to evaluate the interaction between machine learning datasets, territorial jurisdiction, and Section 52 fair dealing exemptions, Indian Express reported.
What this ruling means for AI in India
The Delhi High Court has set an initial legal framework for generative AI operations within Indian jurisdiction while the main suit continues to be heard. The decision marks a crucial precedent for tech developers, media publishers, and IP lawyers across the country, allowing AI companies to train their models on publicly available work.
