Public Interest, copyright and fair dealing: ANI v OpenAI

On 24 July 2026, the Delhi High Court held that Artificial Intelligence (AI) technology is beneficial to society, and that it is in “public interest” to allow AI companies to train on copyrighted material without authorisation. The decision came in a legal action initiated by Asian News International (ANI) against OpenAI, the company behind ChatGPT, in 2024. 

ANI’s lawsuit centred on whether training large language models (LLMs) on copyrighted material constitutes “fair dealing” under the Indian copyright law. The Federation of Indian Publishers, Digital News Publishers Association and the Indian Music Industry also intervened in the lawsuit. 

What is “Fair dealing”? 

While “fair dealing” under Section 52 of the Copyright Act, 1957 is central to Indian copyright law, it remains vaguely defined. Various High Courts have adopted different tests, and no single test has been consistently adopted. The Delhi High Court’s decision in ANI Media v Open AI OPCO (2026) added a new dimension to “fair dealing” by introducing a free-wheeling “public interest” consideration. Without sufficient constraints, it may result in an overly flexible doctrine that is inconsistent with past approaches.

Being a landmark case, there is a possibility that it will reach the doors of the Supreme Court. This may give the Supreme Court an opportunity to settle one of the most important doctrinal questions in Indian copyright law – what are the factors that should go into determining “fair dealing”?

Applicability of “Fair dealing” on AI 

Section 52 lists 33 specific activities that do not require a licence from the copyright-holder. The provision permits these uses because copyright—by design—is not intended to operate as an absolute right. Therefore, the activities listed under Section 52 are generally referred to as “exceptions and limitations” or “user’s rights”. 

Of these 33 activities, one was front and centre in the dispute i.e. Section 52(1)(a)(i). This provision allows for “fair dealing” with any work for the purpose of “private or personal use, including research”. 

The assessment under this provision is two-pronged. 

1. First, is to assess the purpose of the use i.e. whether it is personal or private use or research. 
2. Second, is to assess the “fairness” of the dealing. 

The Single Judge of the Delhi High Court held that OpenAI met both conditions. The purpose test was fulfilled since the training of LLM models was inherently a “private process”. The court relied on the Supreme Court’s judgement in Academy of General Education, Manipal v B. Malini Mallya (2009), which gave an expansive meaning to “private” while holding that a dance performance at an educational institution falls within “private use”. Further, the Court held that AI training qualifies as “research”, which is not limited to human research and extended to machine learning.

The Court then formulated its own three factors for the second condition i.e. “fairness”, in absence of a Supreme Court decision. Broadly, the three factors examine whether the use (1) is limited to a specific purpose, (2) would result in harm to the copyright owner and (3) serves public interest. While factors 1 and 2 can be reconciled with existing approaches to assess fair dealing, factor 3, which is public interest, is a new, independent factor added to the “fairness factors”.

Public interest and copyright law

“Public interest” is an underlying objective that is implicit in the scheme of the Copyright Act.

For instance, Section 52 reflects Parliament’s balancing of public interest—such as education, research, news reporting, and criticism—against the exclusive rights of copyright owners. A Division Bench of the Delhi High Court, in The Chancellor, Masters & Scholars Of University Of Oxford v Rameshwari Photocopy Services (2016), has characterised these activities as “user’s right”, necessary to fulfil the goals of copyright law. Further, the Delhi High Court in The Chancellor, Masters & Scholars Of University Of Oxford v Narendera Publishing (2008) has held that these provisions should be read in the interest of the public in a way that promotes creativity, while guarding against plagiarism. 

The same idea of public interest is reflected in the factors that Indian courts have considered when determining whether a particular use is “fair dealing” under Section 51(1)(a)(i). Courts have considered factors such as – the purpose and character of the use, the nature of the copyrighted work, the proportion of the work used and the effect on the copyright owner’s market. 

These factors come from US copyright law’s “fair use” analysis. While some High Courts have observed that the US four-factor test can be applied as is, others have adapted or modified these factors in their analysis of “fair dealing”. For example, India TV Independent News Service v Yashraj Film (2012), Super Cassettes Industries Ltd. v Chintamani Rao (2011), Syndicate Of The Press Of The University Of Cambridge On Behalf Of The Chancellor, Masters And School v B.D. Bhandari (2011) and Narendera Publishing have endorsed the US four-factor test. Whereas, ESPN Star Sports v Global Broadcast News (2008) used a modified three-factor test. 

Further, High Courts have accorded different weight to different factors within the broader framework. For instance, Super Cassettes Industries Ltd. emphasised proportionality; Narendera Publishing emphasised the transformative character of the use; and India TV emphasised substantiality.

Therefore, High Courts have relied on different factors that mediate between the interests of the copyright holder and that of the public. Public interest is reflected in the cumulative assessment of the various ‘fairness factors’.

It is against this background that the introduction of “public interest” as an independent factor becomes problematic. The Delhi High Court’s approach elevates an underlying objective of copyright law into an independent factor without sufficiently considering its impact. Without meaningful constraints, it allows courts to favour uses that they consider socially desirable at the cost of interests that copyright law itself seeks to protect. 

This question becomes important in the context of generative AI. Whether AI technology is beneficial to society is a broad question of public policy. It involves competing economic, technological, cultural and social interests. It is not necessarily the same question as whether a particular use of copyrighted material is fair under Section 52(1)(a)(i). 

A more principled approach is to recognise that public interest already exists within the statutory scheme and the concept of “fair dealing” itself. Rather than introducing new factors into the determination of fairness, what is needed is a precise set of factors together with the weight to be accorded to each. If the case comes before the Supreme Court, it should examine the factors that have developed over time and formulate a uniform test—one that appropriately mediates between the rights of copyright owners and the interests of the public.

*Aparajita Lath is an IP lawyer and founder of Lath Law (www.lathlaw.com). She can be reached at aparajita@lathlaw.com.