IIPLA News
Thursday, November 19, 2026

Delhi High Court Delivers Landmark Ruling on Copyright and AI Training in ANI v. OpenAI

India’s first major judicial scrutiny of copyright implications in large language model training highlights jurisdiction and fair dealing issues

Delhi High Court Delivers Landmark Ruling on Copyright and AI Training in ANI v. OpenAI

The Delhi High Court’s July 24, 2026 ruling in ANI Media Pvt. Ltd. v. Open AI OPCO LLC [CS(COMM)1028/2024] represents India’s first significant judicial engagement with the intersection of copyright law and large language model (LLM) training. The case, which remains pending on substantive issues, arose from ANI’s allegations that OpenAI unlawfully used its copyrighted news reports to train ChatGPT, resulting in AI-generated responses that reproduced ANI’s content without authorization.

ANI contended that OpenAI’s training of ChatGPT involved unauthorized reproduction of its news reports, infringing its copyright. OpenAI countered by disputing jurisdiction over the training claim, citing its servers’ offshore location, and argued that the training constituted lawful fair dealing under the Copyright Act, 1957. Additionally, OpenAI denied that ChatGPT’s outputs substantially reproduced ANI’s original works.

The Court’s judgment provided an instructive explanation of LLM functioning. It described training as processing vast datasets—including web content and licensed materials—tokenized into units such as words and punctuation. The model learns statistical patterns rather than storing exact copies of training data. The Court distinguished this from Retrieval-Augmented Generation (RAG), where the model actively retrieves external information during response generation.

Four key legal questions framed the Court’s analysis, foremost among them jurisdiction over claims related to offshore servers. OpenAI argued that Indian courts lacked jurisdiction because its servers were located outside India and that ChatGPT’s terms of use designated San Francisco courts for disputes. It also referenced the UK case Getty Images v. Stability AI, where no training claim was pursued.

Rejecting OpenAI’s jurisdictional objection, the Delhi High Court held that it possessed jurisdiction because ANI’s principal place of business and registered office were within its territorial reach, and the allegedly infringing outputs were generated within India for Indian users. The Court characterized the US server storage as a “terminal link” in a continuous chain involving access and transmission of copyrighted works from India and output generation within India.

The Court emphasized that in the era of cloud computing and cross-border data flows, rigid territoriality concepts are inadequate. It cited Neetu Singh & Anr. v. Telegram FZ LLC & Ors. (Delhi High Court, 2022) to support the view that accepting offshore server location as a jurisdictional shield would enable evasion of Indian copyright law.

On the merits of copyright infringement based on ChatGPT’s responses, the Court found ANI had not established a prima facie case. It reiterated that copyright protects expression, not underlying facts or information, and noted that the 14 ANI articles cited postdated the training cut-off dates of the relevant GPT models. The Court considered RAG a more plausible explanation for the AI’s responses than memorization during training, leaving the memorization issue open for trial.

Regarding reproduction through electronic storage, the Court held that both temporary and permanent storage of copyrighted works constitute reproduction under Section 14(a)(i) of the Copyright Act, 1957. The critical question was whether OpenAI’s storage for LLM training qualified as fair dealing under Section 52(1)(a).

The Court contrasted Indian fair dealing with the broader US fair use doctrine, noting that Indian law specifies permitted purposes such as “private or personal use, including research” without enumerating fixed fairness factors. Indian courts have recognized the transformative nature of use as a key consideration.

OpenAI urged a liberal construction of Section 52, arguing AI training is research and private use, not limited to non-commercial contexts. It contended that ‘private’ should extend beyond individual use to closed corporate environments and that ‘research’ need not be human-conducted. OpenAI also emphasized the transformative character of training, asserting ChatGPT is not a substitute for news subscriptions.

ANI countered that commercial AI developers cannot claim private or personal use defenses and challenged the transformative use argument, asserting that raw data’s expressive and non-expressive elements are inseparable in training.

The Court analyzed fair dealing in two stages: purpose and fairness. It rejected ANI’s argument that commercial use is excluded from Section 52(1)(a), noting the Act explicitly limits non-commercial use elsewhere but not in this provision. It distinguished ‘private’ from ‘personal,’ interpreting private use to include internal corporate activities. Since ANI’s content was used internally during training and not publicly distributed, the Court found the use could qualify as private.

Finally, applying principles of statutory interpretation, the Court held that ‘research’ under Section 52(1)(a) need not be human-conducted and can encompass machine learning processes.

This ruling provides critical early guidance on how Indian courts may address copyright issues arising from AI training and use, particularly regarding jurisdictional reach and the scope of fair dealing. The decision underscores the evolving nature of copyright law in the digital and AI era, balancing rights holders’ interests with technological innovation.

The case continues to await trial on substantive issues, with the interim ruling serving as a foundational precedent for future disputes involving AI, copyright, and data use in India.

Share This Article
Ready-to-post copy includes the article link.

Delhi High Court Delivers Landmark Ruling on Copyright and AI Training in ANI v. OpenAI The Delhi High Court’s July 24, 2026 decision in ANI Media Pvt. Ltd. v. Open AI OPCO LLC marks a pivotal moment in Indian intellectual property law by addressing copyright infringement claims related to AI training data... Read the full IIPLA article: https://iipla.org/news/delhi-high-court-delivers-landmark-ruling-on-copyright-and-ai-training-in-ani-v-openai

Related Coverage

Continue in the newsroom

Back to newsroom
PatentsAfrica

Zambian Startup Qualikeeper Secures $75,000 Qualcomm Grant for Livestock AI Platform

Qualikeeper Investments Ltd., a Zambian technology startup, has won a $75,000 grant from Qualcomm’s 2026 Qualcomm for Good Social Impact Fund. The funding supports its AIoT-based livestock intelligence platform, StocKeeper, aimed at improving animal health and strengthening livelihoods in resource-limited farming comm…

Friday, November 20, 2026
PatentsMiddle East

Mothercare Flags Solvency Risks Amid Middle East Franchise Store Closures

Mothercare PLC has issued a cautionary statement on its longer-term solvency following the anticipated closure of most of its Middle Eastern franchise stores, which threatens its financial recovery. The company reported a 14% drop in worldwide retail sales through franchise partners for the half-year ending September…

Friday, November 20, 2026