Generative AI and Copyright Infringement: Deconstructing Fair Dealing Under Section 52 of the Indian Copyright Act
Rohan Vashisht
Partner, Intellectual Property & TMT
West Bengal National University of Juridical Sciences (WBNUJS)
Does scraping publicly accessible copyrighted works for LLM pre-training constitute fair dealing under Indian law? This paper examines Section 52(1)(a) of the Copyright Act, 1957, juxtaposing US fair use doctrines against India’s closed-list statutory exemptions.
1. The Conflict Between Algorithmic Ingestion and Authorial Rights
The exponential growth of Generative Artificial Intelligence (GenAI) models rests upon massive computational scraping of literary, artistic, and musical works. While US courts grapple with *New York Times v. OpenAI* and *Andersen v. Stability AI*, Indian courts face unique statutory constraints under the **Copyright Act, 1957**.
---
2. The Narrow Scope of Indian Fair Dealing
Unlike Section 107 of the US Copyright Act—which provides an open-ended 4-factor test—Section 52 of the Indian Copyright Act is a **closed-list statutory exemption** (*exhaustively defined*).
Under Section 52(1)(a), fair dealing is only permitted for: 1. Private or personal use, including research; 2. Criticism or review, whether of that work or of any other work; 3. The reporting of current events and current affairs.
Commercial machine learning training datasets do **not** neatly qualify as private research or review. Therefore, without a legislative Text and Data Mining (TDM) exception, commercial LLM scraping without licensing creates prima facie copyright infringement exposure under Section 51.
Rohan Vashisht
Partner, Intellectual Property & TMT • West Bengal National University of Juridical Sciences (WBNUJS)
Rohan represents leading AI research labs and media studios in high-stakes patent and copyright litigations.
Share Scholarship
Submit Your Legal Research to Lex Minds
Gain academic visibility, structured editorial feedback, and indexed publication recognition.
Submit Manuscript