Artificial Intelligence and Constitutional Rights: Legal Questions Facing India

Artificial intelligence systems are being deployed across Indian public and private life at a pace that outstrips the legal frameworks designed to govern them. Facial recognition systems operate at airports, railway stations, and in law enforcement applications. Algorithmic tools assess creditworthiness, screen job applicants, and influence decisions about access to government services. Generative AI systems produce content, legal documents, and medical advice. Each of these deployments engages fundamental rights guaranteed by the Constitution of India rights to life and personal liberty, equality, free expression, and due process. This article examines the principal constitutional questions that AI raises in the Indian context, and what those questions mean for businesses, public authorities, and the lawyers who advise them.

Article 21, Dignity and Automated Decision-Making

Article 21 of the Constitution protects the right to life and personal liberty, which the Supreme Court has consistently interpreted to encompass human dignity. The nine-judge bench in Justice K.S. Puttaswamy v. Union of India (2017) 10 SCC 1 identified dignity as an intrinsic element of the right to privacy, itself a component of Article 21. When an AI system makes or substantially influences a decision that affects a person’s life, liberty, or dignity a bail risk assessment, a welfare benefit denial, a credit refusal the constitutional dimension is direct. The person affected may have no knowledge that an algorithm influenced the decision, no ability to interrogate its logic, and no clear avenue of challenge. The absence of an obligation to disclose automated decision-making, explain its basis, or provide human review creates a structural gap between the constitutional guarantee and its practical enforceability. Courts have not yet addressed this gap directly in the AI context, but the analytical framework from Puttaswamy that any state interference with a fundamental right must be proportionate and backed by law will govern such challenges when they arise.

Facial Recognition, Surveillance and Article 21

Facial recognition technology (FRT) is in active deployment by multiple state police forces and at infrastructure controlled by government agencies. The Telangana, Tamil Nadu, Delhi, and Uttar Pradesh police forces have publicly disclosed FRT deployments for investigation and identification purposes. The Aadhaar ecosystem incorporates biometric verification at scale. The legal basis for these deployments is unclear. No specific legislation authorises police use of FRT for surveillance or population-level identification, and the existing frameworks for interception and surveillance the Indian Telegraph Act, 1885 and the IT Act, 2000 do not address real-time biometric identification in public spaces. The proportionality standard from Puttaswamy requires that any interference with privacy be backed by law, pursue a legitimate aim, and use the least restrictive means available. Mass deployment of FRT without a specific statutory basis and without judicial oversight cannot satisfy this standard. Challenges to FRT have been filed before the High Courts, but the constitutional questions have not been conclusively resolved.

Article 14, Equality and Algorithmic Bias

Article 14 guarantees equality before the law and equal protection of the laws. Where an AI system is trained on historical data that reflects pre-existing patterns of discrimination in lending, housing, employment, or policing it may reproduce and amplify those patterns without any human decision-maker consciously applying discriminatory criteria. In the credit market, AI underwriting models trained on datasets where certain caste, regional, or demographic groups are underrepresented as creditworthy borrowers may systematically deny credit to members of those groups. In employment screening, algorithms trained on historically male leadership data may disadvantage women candidates. Where these groups are defined by characteristics protected under Articles 15 and 16 religion, race, caste, sex, place of birth the constitutional dimension is clear. The challenge is evidentiary: establishing that a particular algorithmic system produces discriminatory outcomes requires access to training data, model architecture, and output statistics that are typically treated as proprietary by deploying organisations. No current Indian regulatory framework mandates algorithmic impact assessments or requires disclosure of bias metrics, though the Data Protection Board’s evolving role and sector-specific financial regulators may begin to address this.

Article 19 and AI-Mediated Speech

Article 19(1)(a) guarantees freedom of speech and expression. AI systems now mediate speech in multiple ways: content moderation algorithms decide what online expression is amplified or suppressed; generative AI systems produce speech at scale; deepfake technology creates false attributions of speech to real individuals. Each of these dynamics engages Article 19. Content moderation systems deployed by major platforms in India operate under the 2021 Intermediary Guidelines, which require platforms to take down content within specified timeframes following government orders or court orders. Where an AI moderation system incorrectly suppresses lawful speech a false positive the affected speaker has a constitutional interest in reinstatement and a right to challenge the restriction. The absence of any obligation on platforms to explain why specific content was removed by automated systems, or to provide a meaningful appeal mechanism, creates constitutional friction. Deepfakes that attribute false statements to real persons engage both Article 19 rights of the misrepresented speaker and potential criminal liability under the IT Act and the Bharatiya Nyaya Sanhita, 2023.

AI in the Justice System: Judicial Use and Due Process

The deployment of AI within the justice system itself raises the most direct due process questions. SUVAS and related tools assist in case management and translation; these administrative functions do not engage due process concerns. But if predictive risk assessment tools are used to inform bail decisions, sentencing, or parole recommendations as has occurred in some US jurisdictions the constitutional stakes are higher. Article 21 encompasses the right to a fair hearing before a competent authority, and a decision influenced by an algorithm that the accused cannot examine or challenge may not meet this standard. India has not yet formally adopted risk assessment tools in criminal adjudication, but the trajectory of eCourts development and the enthusiasm for AI-assisted judiciary functions warrants close attention. Any deployment of AI in adjudicative decisions as opposed to administrative case management would require clear statutory authority and robust transparency and challenge mechanisms to be constitutionally defensible.

The Regulatory Landscape: Gaps and Emerging Frameworks

India does not yet have comprehensive AI-specific legislation. The Digital Personal Data Protection Act, 2023 governs the processing of personal data including data used in AI systems, but does not directly address algorithmic accountability, explainability, or automated decision-making rights. The draft National Data Governance Framework Policy addresses public-sector data, not AI deployment standards. The Ministry of Electronics and Information Technology has published advisory documents on responsible AI, but these are not binding. Sector-specific regulators SEBI, RBI, IRDAI have begun issuing guidance on AI use in financial services, which represents the most developed regulatory framework currently in force. In the absence of general AI legislation, the constitutional framework Articles 14, 19, and 21 applied through proportionality analysis provides the primary legal standard against which AI deployments will be assessed by courts. Businesses and public authorities deploying AI should treat constitutional compliance as a live requirement, not a future aspiration.

Accountability Gaps and the Path to Legal Clarity

The most significant challenge in AI constitutional litigation is accountability attribution. Where an AI system produces a harmful outcome, responsibility may be distributed across the developer, the data provider, the deploying organisation, and the human decision-maker who relies on the output. No Indian statute currently resolves this distribution of liability for AI harms. The absence of a clear accountability chain creates risk for every participant in an AI deployment. Organisations that deploy AI systems without maintaining records of training data provenance, model evaluation for bias, and human oversight mechanisms are poorly positioned to defend against constitutional challenges or regulatory action. The practical guidance for now is to document every significant AI deployment decision, maintain explainability records, design human-in-the-loop review for consequential decisions, and monitor the evolving regulatory and judicial landscape closely.