Injunction against ChatGPT would hurt Indian AI development, users: Delhi HC differs with DPIIT white paper

An injunction against OpenAI’s ChatGPT would adversely affect the development of large language models (LLMs) in India and millions of users of the platform, the Delhi High Court held on Friday [ANI Vs OpenAI].

Justice Amit Bansal made the observation while dismissing ANI Media’s application for an interim injunction against OpenAI over the alleged unauthorised use of its news articles to train the LLMs underlying ChatGPT.

The Court said the development and success of LLMs depend on the availability of data in the public domain.

“It would be economically unviable to develop an LLM if training of an LLM would require licenses from multiple sources,” the Court stated.

It added that an injunction would particularly affect LLMs being developed in India.

“Any interim injunction granted at this stage would, in my opinion, be detrimental to the growth of AI and more particularly, to the LLMs being developed in India,” the judge ruled.

The Court also took into account the possible effect on Indian users of ChatGPT, many of whom do not pay for the service.

“It would also have adverse impact on public interest, including millions of users of ChatGPT in India, many of whom would not be paid subscribers,” the order said.

The Court’s reasoning differed from the policy direction suggested by a DPIIT-led committee in its Working Paper on Generative AI and Copyright. The committee had proposed a mandatory blanket licence under which AI developers could use lawfully accessed copyrighted works for training but would be required to pay statutory royalties to rights holders. The working paper did not conclusively decide whether AI training was already protected by fair dealing under the existing law.

The Court, on the facts of ANI’s case, has now held prima facie that OpenAI’s training use falls within Section 52(1)(a) and does not amount to infringement. Its ruling does not establish that licences can never be required for AI training. The finding was based on the private and research-oriented nature of the training process, the absence of substantial reproduction, the lack of demonstrated market harm and the public availability of ANI’s material.

The order was passed on a plea filed by ANI in November 2024, alleging copyright infringement on two counts. First, it claimed that OpenAI had copied and stored ANI’s news content for training its LLMs. Second, it alleged that ChatGPT reproduced its copyrighted articles while generating responses for users.

The case also saw interventions by the Federation of Indian Publishers, Digital News Publishers Association and Indian Music Industry in support of ANI. Flux AI Labs, the Indian Governance and Policy Project and Broadband India Forum also broadly supported OpenAI.

The Court considered whether storage of ANI’s data for training amounted to infringement, whether ChatGPT’s outputs infringed ANI’s copyright, whether OpenAI could claim fair dealing under Section 52 of the Copyright Act and whether Indian courts had jurisdiction when OpenAI’s servers were located in the United States.

The Court accepted that OpenAI stored ANI’s literary works, at least temporarily, during training. Such electronic storage amounted to reproduction under Section 14(a)(i) of the Copyright Act.

However, it held prima facie that the storage was protected as fair dealing for “private or personal use, including research” under Section 52(1)(a).

The word “private” was not limited to an individual and could extend to a company operating within a closed system, the Court held. Since training data was used within the LLM and was not made available to the public for access or download, the use was private.

It further held that the expression “research” must be interpreted in light of advances in technology.

“Research/ learning is no longer confined to humans. It is now being done through Artificial Intelligence,” the Court opined.

It noted that ANI could prevent web crawlers from accessing its website but had not exercised that option. OpenAI, in turn, had stated that it had blocked ANI’s website from its web crawlers for training as well as from ChatGPT’s search and retrieval-augmented generation functions.

The Court also found no material showing that ChatGPT had caused ANI to lose subscribers, advertising income or revenue from its news-syndication business.

Significantly, ANI had offered OpenAI a licence covering its digital media, videos, images, photographs and news content for $7.5 million in October 2024.

The Court said the offer demonstrated that ANI’s claim was quantifiable. If ANI ultimately succeeded in the suit, it could be compensated monetarily. By contrast, an interim injunction could have a significant effect on OpenAI’s functioning that might not be capable of monetary compensation.

The Court said the development and success of LLMs depended on access to data.

It added that requiring licences for material obtained from several different sources would make the development of LLMs commercially impracticable.

Advocates Sidhant Kumar, Akshit Mago, Manyaa Chandok, Anshika Saxena and Lahar Jain from Unum Law appeared for ANI.

OpenAI was represented by Senior Advocate Amit Sibal as well as advocates Sanjeev Kapoor, Nirupam Lodha, Madhav Khosla, Moha Paranjpe, Gautam Wadhwa, Vanshika Thapliyal, Malika Nandkeolyar, Ankit Handa, Darpan Sachdeva, Rajat Bector and Saksham Dhingra.

Senior Advocates Kapil Sibal, Arvind Datar with advocates Shashank Mishra, Akshi Rastogi, Parv Kaushik and Suvaroop Saha Roy appeared for intervenor Broadband India Forum.

Senior Advocate Akhil Sibal with advocates Aditya Gupta, Asavari Jain, Shuvam Bhattacharya, Vani Kaushik, Riddhie Bajaj, Jahnavi Siddhu, Aishwarya Kane and Sauhard Alung appeared for intervenor IGAP Project LLP.

Senior Advocate Haripriya Padmanabhan with advocates Shrutanjaya Bhardwaj, Akshat Agrawal, Tushar Srivastava, Shourya Das Gupta, Siddhi Nagwekar, Yashi Bajpai and Yash Tayal appeared for Flux AI Labs.

Senior Advocates Rajshekhar Rao and Chander Lall with advocates Ameet Datta, Harsh Kaushik, Riddima Sharma, Akshay Nagarajan, Rishikaa, Gauri Khanna and Annanya Mehan appeared for Digital News Publishers Association.

[Read Judgment]