Delhi High Court’s Verdict on ChatGPT: A Boost for AI Development in India
The Delhi High Court has ruled against an interim injunction sought by ANI Media to prevent OpenAI from using its news articles to train ChatGPT. The court’s decision, delivered by Justice Amit Bansal, emphasizes the adverse impact such an injunction would have on the development of large language models (LLMs) in India and the millions of users who rely on ChatGPT.
Justice Bansal highlighted that the growth and success of LLMs heavily rely on the accessibility of publicly available data. He noted that mandating licenses from multiple sources for LLM training would render it economically unviable, particularly affecting Indian-developed models.
“An interim injunction at this juncture would be detrimental to the growth of artificial intelligence, especially LLMs being developed within India,” Justice Bansal stated. He further considered the potential impact on Indian users, many of whom benefit from ChatGPT without subscribing financially.
The court’s stance contrasts with a policy proposal from the Department for Promotion of Industry and Internal Trade (DPIIT), which suggested a mandatory licensing framework for AI developers utilizing copyrighted works, requiring payment of statutory royalties. However, the DPIIT’s working paper did not definitively resolve whether AI training constitutes fair dealing under current copyright laws.
On the specific facts of ANI’s claim, the court found that OpenAI’s use of ANI’s articles for training falls within the scope of Section 52(1)(a) of the Copyright Act, which allows for fair dealing related to private and research use. The court clarified that the training process’s private and research-oriented nature, coupled with the lack of substantial reproduction and demonstrated market harm, supports this determination.
More Read
The case, initiated by ANI in November 2024, alleged copyright infringement based on OpenAI’s storage and use of ANI’s content for training and ChatGPT’s generation of responses derived from ANI’s articles. Various industry and advocacy bodies, including the Federation of Indian Publishers and Broadband India Forum, participated in the proceedings, supporting different sides.
While the court acknowledged that OpenAI stored ANI’s works temporarily, it ruled that this storage falls under fair dealing due to its research nature and limited public access. The court also observed that ANI had not used technical measures to block web crawlers from accessing its content, a point raised by OpenAI, which has since restricted ANI’s site from its training and retrieval functions.
Furthermore, the court found no evidence that ChatGPT’s operations have financially harmed ANI through subscriber loss or reduced advertising and syndication revenue. ANI’s previous offer of a $7.5 million license to OpenAI was cited by the court as evidence of the claim’s quantifiable nature, suggesting monetary compensation would be sufficient if ANI ultimately prevailed.
More Read
The ruling underscores the vital role of data accessibility in LLM development, cautioning against a licensing regime that could hinder innovation. The court’s decision was represented by prominent advocates on both sides, including Sidhant Kumar and Akshit Mago for ANI, and Amit Sibal for OpenAI, among others.
