Supreme Court Asks Government to Consider Petition on High-Risk AI in Welfare Schemes
The Supreme Court of India on Thursday asked the Union government to consider as a representation a petition highlighting the absence of a comprehensive statutory framework for the use of “high-risk” artificial intelligence (AI) systems in public welfare delivery. The petition, filed by advocate N.K. Goswami, raises concerns over the deployment of automated decision-making tools in areas affecting access to food, health benefits, wages, pensions, scholarships and subsidies.
A three-judge Bench led by Chief Justice of India Surya Kant directed the government to treat the detailed petition as a representation and take appropriate action. The Court did not issue a formal notice or admit the matter for detailed hearing, but instead referred it to the executive, noting that the issues raised were largely technical and situated in the policy domain.
The petition questions whether the government can employ AI, machine learning (ML), automated decision-making systems, risk-scoring mechanisms, biometric surveillance, predictive policing technologies and AI-assisted content moderation in critical governance functions without a specific statutory framework or meaningful human oversight. It argues that such systems, when used in welfare-related governance, can directly affect the fundamental rights of citizens, including the right to life, liberty, privacy, dignity, livelihood and free speech.
Mr. Goswami clarified that the petition does not seek a prohibition on technology. “The vice is that technology affecting fundamental rights cannot be permitted to operate outside the discipline of the Constitution. When a citizen’s access to food, health benefits, wages, pension, scholarship, subsidy, liberty, privacy or speech is affected by an opaque automated or semi-automated system, the State must satisfy the minimum constitutional requirements,” he submitted.
The petition emphasised that AI has moved beyond being a mere administrative aid. It pointed to official materials indicating that AI/ML systems, fraud-detection algorithms, beneficiary-verification tools, Aadhaar-linked authentication and data-driven de-duplication are already being used in welfare-linked governance, from health-benefit delivery to ration distribution and fraud prevention. The petitioner argued that these systems often operate without clear legal basis, transparency or grievance redressal mechanisms.
During the hearing, the Chief Justice observed that the Court lacked the necessary technical knowledge to address the policy questions raised. The Bench suggested that the government was better placed to examine the technical and administrative aspects. Mr. Goswami, however, argued that the Court could intervene to fill the legislative vacuum, especially given the absence of any statutory framework governing AI/ML use in public welfare.
The petition has also sought a direction to the Centre to file a complete affidavit disclosing all existing and proposed high-risk AI systems used by Union ministries and agencies in welfare, policing, surveillance and content regulation. It asks for details on the legal basis, vendor or agency involved, purpose, data categories, safeguards, grievance mechanisms and audit status of each system.
The Supreme Court’s decision to refer the matter to the government, rather than take it up judicially, reflects the judiciary’s cautious approach to highly technical policy questions. However, the petitioner’s broader point — that constitutional discipline must apply to all state actions, including those mediated by algorithms — remains a significant legal and democratic concern. The government’s response to the representation will determine whether the issues raised are addressed through policy, legislation, or future litigation.
