The integration of Artificial Intelligence (AI) into governance frameworks is accelerating across the Global South, and India stands at the forefront of this transformation. From biometric welfare systems and predictive policing to algorithmic surveillance, AI is increasingly embedded in public service delivery and state infrastructure. However, this technological expansion occurs within a socio-political landscape deeply shaped by caste, religion, and economic class. This paper critically interrogates how AI systems intersect with India’s entrenched hierarchies, revealing the representational, regulatory, and ethical gaps that threaten to reproduce and entrench structural injustice. Drawing from interdisciplinary frameworks in AI ethics, critical data studies, and postcolonial science and technology studies, the paper engages with concepts such as sociotechnical imaginaries, algorithmic discrimination, and data colonialism. It explores how digital systems often erase class-based identities, resulting in opaque decisionmaking, discriminatory surveillance, and the erosion of privacy and agency for marginalized communities. Through case studies of facial recognition, welfare exclusion, and predictive policing, the paper demonstrates how caste, religious, and economic markers are indirectly encoded into algorithmic governance. The analysis reveals that India's techno-solutionist regulatory model prioritizes innovation and efficiency over rights, accountability, and inclusion. The Digital Personal Data Protection Act, 2023, fails to address algorithmic discrimination, ensure transparency, or mandate oversight. In response, the paper proposes a rights-based, class-conscious AI governance model rooted in India's constitutional commitments to equality, justice, and fraternity. It calls for participatory design, disaggregated data practices, and robust accountability mechanisms to ensure AI serves as a tool of inclusion rather than oppression.