Abstract:

India is embedding artificial intelligence (AI) in public administration at the very moment its legal architecture for data protection, transparency and AI oversight is being rewritten. Between November 2025 and August 2026, the government issued the India AI Governance Guidelines, notified the Digital Personal Data Protection (DPDP) Rules 2025, hosted the AI Impact Summit, and began defending in the Supreme Court a constitutional challenge to the DPDP Act's amendment of the Right to Information (RTI) Act. This paper asks whether administrative efficiency, transparency, accountability and data protection can be reconciled in AI-driven governance in India. Using doctrinal analysis, policy analysis and a case study of Telangana's Samagra Vedika welfare-profiling system, it argues that the four values are not inherently opposed. The tensions arise from design choices, thin institutional capacity and broad exemptions for the State. The paper identifies four structural gaps: voluntary AI guidance without enforceable duties; consent-free State processing and wide exemptions under the DPDP Act; the narrowing of the RTI public-interest test; and the absence of a right to human review and explanation. It proposes a proportionality-based, risk-tiered framework combining statutory authorisation, impact assessments, algorithm registers, meaningful human review, independent audit and accessible redress. Legal positions are stated as at 28 September 2026.

Keywords:

algorithmic governance; e-governance; transparency; accountability; data protection; DPDP Act; proportionality; India.

Aim/Objectives

Methods