The Friction Between Generative AI Training And Data Minimization Under The DPDP Act, 2023

Full Text Link

Vedanti Rajput, Bharatividyapeeth Institute of Management and Research

ABSTRACT

The exponential growth of Generative Artificial Intelligence (AI) has positioned India as a central hub for technological innovation. However, the operational architecture of Large Language Models (LLMs)—which require the ingestion of massive, unfiltered datasets—fundamentally conflicts with the data protection principles established under the Digital Personal Data Protection (DPDP) Act, 2023. This paper examines the systemic friction between the data consumption needs of generative machine learning models and the statutory mandate of "Data Minimization" under Indian law. By analyzing the mechanics of AI ingestion against Section 6 of the DPDP Act, exploring comparative jurisprudence under the GDPR, and evaluating the pitfalls of public data exemptions, this paper demonstrates that the current static regulatory framework creates an unsustainable compliance deadlock. Ultimately, this paper proposes a reconciled regulatory model featuring dynamic safe harbors, synthetic data standards, and AI-specific guidelines to balance technological progress with fundamental privacy rights.

211. THE FRICTION BETWEEN GENERATIVE AI TRAINING AND DATA MINIMIZATION UNDER THE DPDP ACT,

Download THE FRICTION BETWEEN GENERATIVE AI TRAINING AND DATA MINIMIZATION UNDER THE DPDP ACT, • 203KB