Artificial intelligence is increasingly being incorporated into governmental decision-making in areas such as policing, criminal justice, welfare administration, taxation, healthcare, immigration, public employment, education and the delivery of essential services. AI-assisted governance may improve efficiency, consistency and the capacity of public institutions to process large quantities of information. Nevertheless, when algorithmic systems influence decisions affecting liberty, livelihood, privacy, equality or access to public benefits, their use raises serious constitutional and human-rights concerns. Biased datasets may reproduce historical discrimination; opaque models may prevent affected persons from understanding adverse decisions; automated systems may weaken procedural fairness; and fragmented responsibility among government agencies, technology vendors and individual officials may create an accountability deficit. This paper critically examines the human-rights implications of AI-based public decision-making, with particular reference to India and comparative international standards. It argues that ordinary data-protection and administrative-law principles, although important, are insufficient by themselves to address the distinctive risks of algorithmic governance. India requires a rights-based legal framework governing high-risk public-sector AI. Such a framework should require lawful authority, necessity, proportionality, algorithmic impact assessments, equality audits, meaningful human oversight, public registers, reasoned decisions, independent supervision and accessible remedies. AI should support public authorities but should not become an unreviewable substitute for constitutionally accountable human judgment.