الهند

This is an informal case summary prepared for the purposes of facilitating exchange during the 2026 WIPO IP Judges Forum.

Session 3: Copyright and AI Model Training

High Court of Delhi, India [2026]: ANI Media Pvt. Ltd. v. Open AI OpCo LLC, Case No. CS(COMM) 1028/2024, I.A. 45300/2024

Date of judgment: July 24, 2026 (Corr. July 27, 2026)

Issuing authority: High Court of Delhi

Level of the issuing authority: First Instance

Type of procedure: Judicial (Civil)

Subject matter: Copyright and Related Rights (Neighboring Rights)

Plaintiff/Appellant: ANI Media Pvt. Ltd. 

Defendant/Respondent: Open AI OpCo LLC

Keywords: Artificial intelligence (AI), Large language models, Generative AI, ChatGPT, Copyright infringement, Reproduction right, Text and data mining, Training data, Fair dealing, Retrieval-augmented generation, Interim injunction, Balance of convenience, Extraterritorial jurisdiction

Basic facts: The plaintiff, ANI Media Pvt. Ltd. (ANI), a prominent Indian news agency sued the defendant, Open AI OpCo LLC (Open AI), the developer of the generative AI chatbot ChatGPT, alleging unauthorized use of ANI's copyrighted literary works (news content).

ANI's case rested on two claims: (i) a "training claim" – that Open AI copied and stored ANI's works without licence, to train the large language models (LLMs) underlying ChatGPT; and (ii) a "reproduction/output claim" – that ChatGPT reproduced ANI's copyrighted content, including near-verbatim extracts, in its responses to users.

ANI filed an interlocutory application (I.A. 45300/2024) under Order XXXIX, Rules 1 and 2 of the Code of Civil Procedure, 1908, seeking an interim injunction restraining Open AI from using ANI's works to train its LLMs and from generating outputs reproducing or paraphrasing such works, besides other reliefs including the takedown of already-stored data.

Open AI resisted the application, contending, inter alia, that its use of publicly available data for training LLMs constituted "fair use"/fair dealing, that ChatGPT's outputs were not substantially similar to ANI's works, that the Indian courts lacked jurisdiction since training occurred on servers located outside India, and that ANI could have used technical opt-out/crawler-blocking tools to prevent scraping of its website but chose not to do so.

Several third parties intervened: the Federation of Indian Publishers, the Digital News Publishers Association, and the Indian Music Industry broadly supported ANI, while Flux AI Labs, IGAP Project LLP and the Broadband India Forum broadly supported Open AI; two Amici Curiae were also appointed to assist the Court.

On 19 November 2024, the Court framed four issues for adjudication of the interim injunction application:

      I.        Whether Open AI's storage of ANI's data for training ChatGPT amounted to copyright infringement;

    II.        Whether use of ANI's data to generate responses for users amounted to infringement;

   III.        Whether Open AI's use qualified as "fair use"/fair dealing under Section 52 of the Copyright Act, 1957; and

  IV.        Whether Indian courts had jurisdiction given that Open AI's training servers are located in the United States of America.

Held:  The Court dismissed ANI's application for interim injunction.

On jurisdiction (Issue IV), it held, on a prima facie basis, that Indian courts had territorial jurisdiction notwithstanding that Open AI's training servers were located outside India, since the training and output claims could not be severed and ChatGPT's outputs were generated for and received by users within the jurisdiction.

On the reproduction/output claim (Issue II), the Court held that ANI had failed to establish, at the prima facie stage, that ChatGPT memorized and regurgitated ANI's works or that its outputs were a substantial reproduction of ANI's news articles; the illustrations relied upon by ANI post-dated the cut-off dates of the LLMs' training data.

On the training/storage claim and the fair-use/dealing defense (Issues I and III, considered together), the Court held that Open AI's storage of ANI's literary works for training its LLMs fell within the fair-dealing exception under Section 52(1)(a)(i) of the Copyright Act, 1957 ("private or personal use, including research"), since both the "purpose test" and the "fairness test" were satisfied; the commercial nature of Open AI's business did not disentitle it from this defense, as Section 52(1)(a), unlike certain other clauses of Section 52(1), does not expressly exclude commercial use.

On balance of convenience and irreparable injury, the Court found that ANI's claim was quantifiable in monetary terms (ANI having earlier offered Open AI a license for USD 7.5 million), that ANI had not availed of technical opt-out/crawler-blocking tools available to it, and that the public interest in the continued availability of ChatGPT to millions of Indian users and in the growth of AI/LLM development in India weighed against grant of an injunction.

Accordingly, the case was dismissed, with the Court clarifying that its observations were confined to the interim stage and would have no bearing on the final outcome of the suit, which remains pending.

Relevant holdings in relation to Artificial Intelligence Model Training and Copyright: Storage/copying of copyrighted literary works by an AI developer for the purpose of training an LLM constitutes reproduction/storage within the meaning of Section 14(a)(i) of the Copyright Act, 1957, but such storage, where confined to training, can fall within the fair-dealing exception for “private or personal use, including research” under Section 52(1)(a)(i).

The commercial character of the defendant’s enterprise does not, by itself, take the use outside Section 52(1)(a), since the Act expressly restricts other specific clauses of Section 52(1) to non-commercial use but contains no such restriction in Section 52(1)(a).

The “non-infringing copy” requirement in the Explanation to Section 52(1)(a) that storage of an infringing copy of a computer program would not be protected even if stored incidentally, applies only to computer programs and not to literary or other works stored electronically.

Whether the use qualifies as fair dealing depends on a “purpose test” and a “fairness test”, the latter examining, among other things, whether the use is confined to training (as opposed to independent commercial exploitation of the works themselves), whether it results in economic substitution/competition prejudicing the copyright owner’s legitimate interests, and whether the AI tool serves a broader public interest.

An output/reproduction claim against a generative AI tool requires proof, at least prima facie, of memorization and substantial (qualitative and quantitative) reproduction of the claimant’s specific expression in the actual outputs generated; training on copyrighted works, by itself, does not establish that outputs are infringing.

The availability of technical opt-out or crawler-blocking mechanisms for copyright owners, and whether such mechanisms were used, is a relevant consideration in assessing balance of convenience in AI-copyright disputes. Public interest in the continued availability and development of AI/LLM technology, including for users and developers in India, is a material factor under the fourth prong of the interim-injunction test in intellectual property matters.

Relevant legislation: Copyright Act, 1957 (IN153); Code of Civil Procedure, 1908 (IN159)