To systematically map and critically analyze the provisions within Indian national frameworks (Digital Personal Data Protection Act 2023, ICMR guidelines) and key international ethical AI guidelines that address privacy and data security for AI-driven diagnostic tools, specifically within the Indian public health context. A scoping review was conducted following the PRISMA-ScR framework. A comprehensive search of government publications, intergovernmental organization reports, and academic literature was performed to identify relevant national and international guidelines. Data were charted and synthesized thematically. The analysis reveals a fragmented Indian governance landscape, characterized by a principles-based national AI strategy (#AIforAll), sector-specific ethical guidelines (ICMR), and a new, general-purpose data protection law (DPDP Act). While the DPDP Act establishes foundational data fiduciary obligations, its broad exemptions for public health and research create ambiguities for data-intensive AI. International frameworks serve as comparative reference points; the EU AI Act, in particular, offers a contrasting granular, risk-based regulatory model. Key tensions emerge between the universalist principles of international guidelines and the need for a “situated ethics” approach that addresses India’s unique challenges, including the digital divide, data quality issues, and the difficulty of operationalizing meaningful consent. Existing guidelines provide a foundational but incomplete framework for the safe and ethical deployment of AI diagnostics in Indian public health. There is a critical need to bridge the gap between high-level principles and on-the-ground implementation. This requires developing sector-specific regulations under the DPDP Act, establishing robust standards for data and algorithmic audits, and fostering a cooperative federalism approach that harmonizes international standards with the socio-technical and constitutional realities of India.
Joshi et al. (Tue,) studied this question.