India - ANI Media Pvt Ltd V. OpenAI OPCO LLC, 24 July 2025

Claimant: ANI Media Pvt Ltd 

Defendant: OpenAI OPCO LLC 

Both parties have filed their opening arguments. No interim injunction was granted against OpenAI’s operations; the DHC observed that immediate harm was not evident since OpenAI stopped using ANI content by removing it from its web crawlers. 

On 4 July 2025, the DHC disposed of various intervention applications, permitting the intervenors to argue only on legal issues. The Claimant's counsel presented arguments in rejoinder. Subsequent hearings in July, August and September 2025 focused on submissions from the intervenors. 

On 27 March 2026, the DHC reserved the judgment, after the conclusion of arguments subsequent to 32 individual hearings before Justice Amit Bansal, with six intervenors primarily participating: three supporting ANI — DNPA, Indian Music Industry, and Federation of Indian Publishers; and three supporting OpenAI — Broadband India Forum, Flux Labs, and IGPP.

On 24 July 2026, the DHC passed an interim order refusing ANI’s request for an interim injunction, holding that at the prima facie stage, OpenAI’s use of ANI’s content to train ChatGPT models falls under the ‘fair dealing’ exception under Section 52(1)(a)(i) of the Copyright Act.

1. Asian News International (“ANI”), is a leading multimedia news agency with over 100 bureaus in India and across the globe. ANI’s  range of products include edited news feeds and customized programmes for television channels, audio bytes for radio stations, live web casting and streamed multimedia / text content for websites and mobile carriers, and news wire services for newspapers, magazines, and websites.
2. ANI is seeking to restrain Open AI from using ANI’s copyright materials to train its AI models.  
3. ANI sought in October 2024 to block its s domain i.e. ‘www.aninews.in’ from being used to train  Open AI’s systems
4. The primary issues pending before DHC are as follows:
   1. Whether the storage of ANIs data for training Open AI’s ChatGPT system amounts to infringement? ;
   2. Whether the use of data to generate responses for users of ChatGPT amount to infringement?
   3. Whether Open AI’s use of ANI copyrighted materials qualifies as Fair Use under S. 52 of the Indian Copyright Act, 1957 ? ;
   4. Whether Indian Courts have jurisdiction for the present lawsuit? 
5. The DHC appointed 2 Amici Curiae owing to the range of issues involving varied technological advancements: 1) Practicing Lawyer in the field of IPR, including Copyright; 2) Academician in the same field. Accordingly, Mr. Adarsh Ramanujan, Advocate and Dr. Arul George Scaria, Professor of Law, National Law School of India University (NLSIU) were appointed to assist the Court.
6. Given the potential precedent-setting nature of case, numerous other stakeholders sought to intervene in the proceedings on one side or the other.
   1. The Digital News Publishers Association (DNPA) – an industry body representing major Indian news outlets, Federation of Indian Publishers (FIP), Indian Music Industry (IMI) filed intervention applications to join the ongoing suit and express their respective  concerns from the use of data by Open AI. 
   2. An independent think-tank, the Indian Governance and Policy Project (IGPP) sought to offer views on broader policy implications of AI and copyright. 
   3. On the other side, Flux Labs, an Indian AI startup, intervened to support OpenAI’s stance and argued that merely extracting non-expressive elements of data for training should not count as infringement.
7. On 24 July 2026, the DHC delivered its first substantive ruling in the matter, declining to grant ANI interim relief and holding that OpenAI’s training-related use of ANI’s content is, at the prima facie stage, protected by the fair dealing exception; the main suit continues to be heard on merits.

1. ANI sought an injunction restraining Open AI, or any person acting on their behalf, either directly or indirectly, from storing, publishing, reproducing or in any manner using, including through ChatGPT model, the copyrighted works of ANI.
2. It has also sought an order from the Court to direct Open AI to disable access of ChatGPT to ANI’s published works.
3. By way of its order dated 24 July 2026, the DHC declined to grant interim relief in line with remedy (i) above; the main suit, including remedy (ii), remains pending on merits.

ANI’s Claims

1. ANI claims that Open AI is using its copyrighted material without permission to train its AI models.
2. ANI claim that ChatGPT gives false information as output and cites the ANI’s news agency as its source, thereby posing a threat to its reputation and leading to the spread of misinformation in the digital sphere.
3. While some of ANI’s content is publicly available content, this does not grant to Open AI the right to exploit or copy its copyright works. 

OpenAI’s Defence

1. Jurisdiction: Open AI claims that the DHC’s Commercial Division lacks jurisdiction, as it combines multiple causes of action, some of which fall outside its competence as for “commercial disputes” as defined under the Commercial Courts Act, 2015..”
2. Open AI claims it is not responsible for the operation of the services provided by ChatGPT, nor is it the owner or lessee of the servers on which the training data is stored. Open AI further claim that none of the training of its AI model takes place in India
3. Open AI claims that Indian Copyright law protects expression and not underlying facts. Since the models do not store or provide identical reproductions in the input materials, it does not infringe copyright. Open AI claim they have only used freely available information to provide its analytic processes and that it has no purpose or intent to coy ANI materials.
4. Open AI contends that the output produced by ChatGPT is protected under “fair use”- “transformative purposes, such as machine training, are not substantial reproduction of original work itself.”  1 Further Open AI argues that ANI’s copyright works constitutes factual content which is not capable of protection as copyright works.
5. During the course of the oral arguments heard on 22 April 2025, the DHC questioned whether excerpts from an exclusive interview can be reproduced by someone else. The question was posed to understand whether any entity has the right to reproduce, through different words and expressions (or by mere paraphrasing), an interview that was conducted by someone with their expenses and resources. Open AI’s counsel answered that interviewers do not get ownership or authorship of facts/ideas disclosed by an interviewee. It was further argued that an exclusive interview will be solely owned by the speakers/interviewees as there is no originality on the part of the journalist.
6. Open AI has also submitted that ANI has attempted to manipulate the model by providing deliberate prompts containing excerpts of their news articles, thereby attempting to force it to recreate the content. However, as ChatGPT presents different responses based on patterns, it demonstrated its ability to generate original words. They submitted that regurgitation is almost an impossible outcome and frowned upon in the concept of LLMs.
7. Open AI also submitted that “hallucinations” of ChatGPT are within the limitations of the model and do not constitute actual harm to the news agency’s reputation, thereby negating the claims of defamation.
8. Open AI additionally submitted that there is no allegation or evidence of commercial harm, or loss of revenue suffered by ANI and that ChatGPT drives traffic to ANI's website. Open AI further clarified that ChatGPT has various purposes, and it is not a competitor to ANI in the news service.
9. At final arguments, Open AI further submitted that Indian copyright law does not prohibit the mere use of data, that AI training is analogous to reading or learning from material, and that it maintains an opt-out blocklist for publishers who do not wish their sites to be used for training — noting that it had blocked ANI’s site from further training since October 2024.

ANI’s Response to Open AI’s arguments:

On 4 July 2025, arguments in rejoinder were presented by ANI where it argued that:

1. On whether news items are capable of holding copyright, the counsel for ANI relied on the principle established in the RG Anand v. M/S Delux Films & Ors case wherein it was held that ‘copyright subsists in the expression of an idea, not in the idea itself’. ANI contended that if this distinction is not maintained, it would effectively exclude all non-fiction works (including journalism, legal writing, historical narratives, etc.) from copyright protection, which cannot be the law. It argued that merely because a work is in the nature of a news report, it does not automatically exclude it from limited copyright protection. The factual content may be in the public domain, but the manner of presentation, narrative structure, and original selection of facts can still be protected. 
2. In response to OpenAI's argument that ANI is not the first owner of copyright under Section 17 of the Copyright Act, 1957, ANI submitted several commercial agreements, characterized as employment or professional contracts, to demonstrate its ownership of the interviews. These included documentation showing that the interviews were conducted by ANI’s Chief Editor, an employee of the organization. ANI also cited specific contractual clauses to establish that the works were either created under a contract of service or were expressly assigned to ANI. Additionally, several affidavits were submitted by the authors themselves, affirming that ANI is indeed the first owner of the copyright in question.
3. In response to OpenAI’s reliance on the fair use exception under Section 52(1)(a) of the Copyright Act, 1957 (specifically, the clause relating to “private and personal use including research”), ANI argued that such use does not fall within the scope of this provision. ANI contended that the mere fact that content is publicly accessible on its website does not amount to a license for unrestricted use. They further asserted that the defense of "public availability" applies only when the copyright has either expired or has been expressly relinquished by the author. Moreover, ANI argued that where the predominant purpose of use is commercial, the “research” exception cannot be invoked. Since OpenAI operates on a subscription-based model and uses the content for commercial purposes, ANI claimed that it cannot avail itself of the “research” defence.

Submissions of the Intervenors

A. Digital News Publishers Association (DNPA) and The Indian Music Industry (IMI).

1. In early August 2025, it was argued that the unlicensed "scraping, storing, and reproducing" of news articles by AI models threatens the survival of journalism by reducing the incentive to create content. A warning was issued by the counsel for DNPA that "Physical newspapers are disappearing, digital news will disappear, and only ChatGPT will remain".
2. The submission was made that infringement occurs the moment content is downloaded without permission, stating, "Even storing an infringing copy for a transient moment amounts to infringement".
3. It was argued that India's "fair dealing" is a closed list of specific exceptions, unlike the more open-ended "fair use" doctrine in the U.S.. The argument followed that OpenAI does not qualify for any of the specific exceptions for reproduction under Indian law.
4. In response to the defence that facts cannot be copyrighted, it was clarified that while facts themselves are not protected, the "expression of those facts" is what receives copyright protection.

B. AI-based shopping platform Zoop (Flux Labs)

1. It was submitted that AI tools like ChatGPT synthesize and compile publicly available information and do not reproduce works verbatim.
2. The argument was made that the outputs are closer to derivative works and that "there is neither reproduction of the whole work nor any substantial part thereof, and mere storage per se does not constitute infringement". 

The Court’s Findings (24 July 2026)

By way of order dated 24 July 2026 (“Order”), Justice Amit Bansal dismissed the interim application adjudicating the pending issues before the DHC. In the Order, the DHC reasoned as follows:

1. Whether Indian Courts have jurisdiction for the present lawsuit?
   • The DHC held that, insofar as territorial jurisdiction is concerned, it was an admitted position that ANI’s principal place of business and registered office are situated within the jurisdiction of the DHC. Further, the DHC noted that OpenAI specifically targets and offers its services to users and subscribers across India, including those located within the jurisdiction of the DHC.
   • Additionally, the alleged infringing activities pleaded in the plaint, including the generation of responses by OpenAI in India based on prompts provided by ANI, were stated to have occurred within the jurisdiction of the DHC. Accordingly, at the prima facie stage, the DHC held that sufficient grounds existed to establish its territorial jurisdiction to entertain the present suit.
2. Whether the use of data to generate responses for users of ChatGPT amounts to infringement
   • The DHC noted that OpenAI did not dispute that ANI’s news articles and interviews qualified as “original literary works”. On the issue of ownership, DHC observed, prima facie, that the Professional Services Agreement placed on record expressly recognised ANI as the owner of copyright in original literary works created by its personnel, in accordance with Section 17 of the Copyright Act.
   • With respect to memorisation, the DHC observed that an LLM is not designed to reproduce training data verbatim, but instead generates outputs by predicting text based on learned patterns. DHC noted that memorisation may arise only in limited circumstances and held, prima facie, that the responses in question could not have resulted from memorisation of ANI’s works during training but rather using the RAG technique, through which LLMs retrieve relevant information from external sources. 
   • With respect to the concern of ‘substantial similarity’, the DHC reiterated the settled position that copyright protection does not extend to facts themselves, but only to the particular form and manner in which those facts are expressed. Examining the examples relied upon by ANI, including the Neeraj Chopra interview, the DHC found that ChatGPT’s responses did not constitute substantial reproductions of ANI’s articles, but instead reflected independent expression, commentary, and contextualisation.
3. Whether the storage of ANIs data for training Open AI’s ChatGPT system amounts to infringement? and Whether Open AI’s use of ANI copyrighted materials qualifies as Fair Use under S. 52 of the Indian Copyright Act, 1957 ?
   • The DHC noted that the above issues were intertwined and had to be considered together. The DHC rejected ANI’s argument that “private or personal use, including research” under Section 52(1)(a)(i) is limited to non-commercial activities or individual users. It held that the provision does not contain any such restriction and that “private” can include use by a private entity where the data remains within a closed space without being made available to public. 
   • The DHC further observed that there is no single test for determining fair dealing and formulated a three-factor assessment for the present case: (i) whether the use of ANI’s works was limited to LLM training; (ii) whether such use resulted in market competition or prejudice to ANI’s legitimate commercial interests; and (iii) whether ChatGPT’s functions served a broader public interest.
   • Applying this test, the DHC found, prima facie, that ANI had not established that OpenAI used its works for purposes beyond LLM training or that ChatGPT’s outputs substituted ANI’s news content. It noted that ChatGPT serves a different function from ANI’s news reporting and syndication business, and that ANI had not provided evidence of any loss of market share or revenue. Recognising the broader benefits of LLMs in areas such as research, education, accessibility and technological development, the DHC held that the public interest factor was also satisfied. Accordingly, the DHC concluded, prima facie, that OpenAI’s use of ANI’s works for training its LLMs constituted fair dealing and did not amount to copyright infringement.