Automated Decision-Making By The State In India - Constitutional Accountability Under Articles 14 And 21
Author: LawSphereX
Category: Constitutional & Human Rights
Published:
Updated:
AUTOMATED DECISION-MAKING BY THE STATE IN
INDIA: CONSTITUTIONAL ACCOUNTABILITY UNDER ARTICLES 14 AND 21
Pernica Shankar, Vivekananda Institute of Professional Studies (IPU, Delhi)
ABSTRACT
The use of technology in public administration is increasing in India, and more government decisions are being made with the help of algorithms. While these systems promise efficiency and objectivity, they also raise important questions about transparency and accountability. Many algorithms function as “black boxes,” and it is often unclear how decisions are made or who is responsible if something goes wrong. This paper looks at the constitutional challenges of algorithmic decision-making in India. By examining Articles 14 and 21, and key court cases like Puttaswamy and Maneka Gandhi, it considers how algorithms may affect equality, privacy, and personal liberty. The paper argues that without proper legal safeguards, automated systems risk reinforcing biases and violating citizens’ rights.
A stronger focus on transparency, human oversight, and a “technological due process” is necessary to ensure these tools protect rather than harm fundamental freedoms. Keywords: Automated Decision-Making, State Accountability, Article 14, Article 21, Rule of Law.
Introduction
India is rapidly adopting technology in governance. Today, decisions in welfare, law enforcement, and even the judiciary increasingly involve automated systems. For example, Aadhaar-linked welfare programs rely on biometric data to deliver benefits, and the Supreme Court has piloted SUPACE, a tool that helps judges with legal research. While these systems can make administration faster and reduce human errors, they are not completely neutral. Algorithms are built on data, and that data often contains biases—related to caste, gender, or socio-economic status. If not carefully managed, these systems can unintentionally harm citizens’ rights. Currently, India lacks clear legal rules to regulate the use of algorithms by the State. Existing laws, like the IT Act, offer only partial safeguards.
They do not fully address problems like automated discrimination, opaque decision-making, or the lack of human accountability. This paper explores the legal and constitutional issues raised by algorithmic governance and argues for a rights-based approach that strengthens citizens’ liberty and equality. 1 Research Objectives The primary objectives of this study are: • To understand how the State uses algorithms for decision-making. • To explore how automated systems may affect fundamental rights in India. • To examine issues of accountability and transparency in algorithmic governance. • To suggest possible legal measures that could help prevent unfair outcomes caused by these systems. 1. Justice K. S. Puttaswamy (Retd. ) v. Union of India, (2017) 10 SCC 1.
[This landmark ruling serves as the primary anchor for this paper, as it shifted the burden of proof onto the State to justify any digital intrusion into the "inner recesses" of the human personality. ] 2. Gurneet Singh Chawla, "Algorithmic Bias in Indian Criminal Justice System: A Constitutional Challenge," IJALR. [Chawla specifically notes that AI assistance in law enforcement is not merely a technical update but a transformative shift in the exercise of State power. ] 3. Maneka Gandhi v. Union of India, (1978) 1 SCC 248. [Cited here to illustrate the foundational requirement that all Statemandated procedures—including those automated—must be "just, fair, and reasonable. " Any "black box" algorithm that fails this test is essentially unconstitutional under the Gandhi standard. ]
Research Methodology
This paper uses a doctrinal research approach. It analyses constitutional provisions, mainly Articles 14 and 21, to see how they apply to automated governance. Judicial decisions like Puttaswamy and Maneka Gandhi are studied to understand the courts’ approach to rights and fairness. Secondary sources, such as legal journals, government reports (from NITI Aayog and MeitY), and international frameworks like the EU AI Act, are also reviewed to identify global best practices and gaps in Indian regulation. Concept of Algorithmic Decision-Making An algorithm is basically a system that can make predictions or decisions based on data. Some algorithms are fully automated, meaning the entire process—from collecting information to reaching a decision—can happen without human involvement. The State uses such systems in areas like welfare distribution, law enforcement, and judicial assistance.
Algorithmic Governance in India Algorithmic governance is already part of daily life in India. For example: • The Aadhaar-linked Public Distribution System (PDS) uses fingerprints to verify beneficiaries. Sometimes errors or mismatches prevent people from getting essential food supplies. • The SUPACE system is being used by courts to help with case research, though it does not make final judgments. • Facial Recognition Systems (FRS) have been used by law enforcement, such as during protests in Delhi, often without clear laws or publicly disclosed accuracy rates. Despite this widespread use, most of these systems operate without audits, legislative approval, or public transparency. This raises concerns under administrative law, which requires government action to be legal, reasonable, and subject to judicial review. 2 4. E. P. Royappa v. State of Tamil Nadu, (1974) 4 SCC 3.
[This case birthed the "new doctrine" of equality. For algorithmic governance, this means that even if a system is mathematically consistent, it may still be "arbitrary" if its logic cannot be explained or justified to the citizen it affects. ]
Constitutional Implications Article 14 – Equality and Arbitrariness : The risk of biased algorithms poses a direct threat to the constitutional guarantee of equality before the law under Article 14. Algorithmic systems trained on historical data often perpetuate structural inequalities related to caste, gender, or religion. This leads to algorithmic discrimination, where individuals are unfairly disadvantaged because the mathematical model replicates pre-existing societal prejudices. Furthermore, the "black box" nature of AI leads to a lack of transparency, making it impossible for citizens to know how a decision was reached. In Indian jurisprudence, equality is the antithesis of arbitrariness. As held in E. P. Royappa, an arbitrary act is a violation of Article 14; thus, any automated decision that lacks a rational, explainable basis is a violation of equality before the law.
Article 21 – Due Process and Dignity : The deployment of algorithms in governance has a profound impact on personal liberty and individual autonomy. Under Article 21, the State cannot deprive a person of life or liberty except through a procedure that is "fair, just and reasonable". The absence of human reasoning in ADM challenges this "technological due process," as automated suspicion and profiling can dilute the traditional requirement of a "reason to believe" in criminal investigations. The right to live with dignity is intrinsic to Article 21, as established in Maneka Gandhi v. Union of India, which expanded the clause to include substantive due process. Furthermore, Justice K. S. Puttaswamy v. Union of India recognized privacy as a fundamental right, emphasizing that data processing by the state must meet tests of legality, necessity, and proportionality.
A critical missing component in India is a statutory right to explanation, which would mandate that individuals be informed of the logic behind automated decisions that significantly affect them, a protection currently lacking compared to international models like the GDPR. Issue of Accountability and Transparency This section addresses the core analytical challenge: the "accountability gap" in algorithmic systems. 5. Shreya Tiwari, "Towards Rights-Based AI Framework In India," LiveLaw (2025). [Tiwari argues that Article 19(1)(a)'s right to information should be extended to include a "Right to Explanation" for any AI output affecting a citizen. ] 6. Justice K. S. Puttaswamy (Retd. ) v. Union of India, (2018) 1 SCC 1 (Aadhaar Judgment). [ Here the Court warned that while efficiency is a legitimate goal, it cannot "steam-roll" fundamental freedoms. ]
7. Cathy O'Neil, Weapons of Math Destruction
Responsibility for Errors: A primary sub-point is who is responsible for algorithmic errors? Many public service AI systems are developed by private vendors under opaque contracts, creating a legal vacuum where it is unclear if liability rests with the vendor, the government agency, or the algorithm itself. This violates the constitutional principle that executive action must be accountable. • Difficulty in Challenge: There is a significant difficulty in challenging automated decisions because citizens often lack the technical and financial resources to contest complex code. Current grievance mechanisms, like consumer courts, are not equipped to manage the technical nuance of "algorithmic harm".
3 • Absence of Explainability: The absence of explainability means that even if a person is denied a loan or profiled by police software, they have no legal mechanism to seek corrections or redress. Transparency is a prerequisite for accountability; without knowing the "logic involved," the right to a remedy remains illusory. • Rule of Law Concerns: From a Rule of Law perspective, algorithmic governance threatens to invert accountability. In a democratic society, the State should be transparent to the citizen, but algorithms often render the citizen transparent to the State while the State's decision-making remains opaque. This shift toward a "nanny state" that monitors every individual interaction without clear limits is anathema to the concept of limited government. To remedy these issues, India must move from voluntary guidelines to binding standards for non-discrimination.
This includes mandatory algorithmic impact assessments and the establishment of independent regulatory bodies with the technical capacity to audit AI systems and ensure constitutional compliance. State Justifications and the Limits of Judicial Intervention The State often justifies the deployment of algorithmic systems on grounds of administrative 8. Anuradha Bhasin v. Union of India, (2020) 3 SCC 637. [This case established that the use of technology for surveillance/restrictions must be proportionate. This paper argues that this principle should be applied ex-ante to any algorithm before it is deployed against the public. ] 9. Lilian Edwards & Michael Veale, "Slave to the Algorithm? " (2017). [The authors suggest that merely providing the "source code" isn't enough; true transparency requires an "explanation of the black box" in terms an ordinary citizen can understand. ] 10.
EU AI Act, Regulation (EU) 2024/1689. [This serves as a vital comparative model, particularly its Article 5, which prohibits "unacceptable risk" AI practices like social scoring. ] 11. Draft Personal Data Protection Bill, 2018 (Srikrishna Committee Report). [While focused on data, the report emphasizes that data "principals" (citizens) must retain ownership over their information, a concept that must be ported into algorithmic governance. ]
efficiency, reduction of corruption, and the need to govern at scale in a populous democracy. Automated systems are presented as tools to eliminate human discretion, standardize decisionmaking, and minimize welfare leakages. Courts have traditionally shown deference to executive policy choices in matters involving technology and governance, particularly where such measures are framed as facilitating effective administration. However, efficiency cannot operate as a constitutional trump card. As repeatedly affirmed by the Supreme Court, administrative convenience cannot override fundamental rights. Judicial deference must yield where opaque systems produce disproportionate or discriminatory outcomes.
Algorithms, like all instruments of State power, must remain subject to constitutional scrutiny, especially when their operation results in exclusion, surveillance, or deprivation of basic entitlements. Challenges and Legal Gaps in India The current landscape of algorithmic governance in India is marked by several critical gaps that threaten to undermine the protection of fundamental rights: • No Specific AI Legislation: India currently lacks a comprehensive legal framework specifically regulating artificial intelligence or state-used algorithms. While the Digital Personal Data Protection (DPDP) Act, 2023 covers some data aspects, it fails to address AIspecific issues like black-box decision-making and algorithmic discrimination. • Limited Judicial Guidance: Indian jurisprudence regarding AI and its interaction with the Constitution is still in its nascent stages.
While courts have addressed internet shutdowns and general data privacy, they have yet to establish a clear doctrinal framework for algorithmic accountability. • Lack of Procedural Safeguards: There is a glaring absence of a statutory "right to explanation" in India. Unlike traditional administrative actions, algorithmic outputs often lack the procedural fairness of a notice or hearing, which are core requirements for a "just, fair, and reasonable" procedure under Article 21. • Digital Illiteracy: A large portion of the Indian populace remains illiterate and poor, creating a significant digital divide. This makes it nearly impossible for affected citizens to understand or contest complex algorithmic decisions that might exclude them from essential welfare benefits.
Comparative Insight: The EU Approach The European Union has pioneered a risk-based regulatory approach through the EU AI Act and the GDPR. Under the EU AI Act, systems are classified into four risk levels: unacceptable, high, transparency, and minimal risk. A landmark feature of this regime is the "Right to Explanation" found in Article 86 of the AI Act and Recital 71 of the GDPR. This grants affected individuals the right to obtain clear and meaningful explanations for decisions made by highrisk AI systems that produce legal or significant effects. Furthermore, the EU framework mandates strict safeguards for high-risk systems, including mandatory human oversight, detailed documentation, and activity logging to ensure results can be traced and audited. This highlights the importance of a framework that prioritizes transparency and fairness over technical efficiency alone.
It is important to note that the “Right to Explanation” under European law is not absolute and remains subject to scholarly debate regarding its scope and enforceability. Nevertheless, the EU framework reflects a clear normative commitment to transparency, auditability, and human oversight—principles that remain conspicuously absent in the Indian regulatory landscape. Suggestions and Safeguards To ensure that technology serves human dignity rather than eroding it, India should adopt a rights-based AI framework: • Transparency in Systems: Developers should be under a statutory obligation to design interpretable systems. This includes disclosing the data sources and the logic used in decisionmaking. • Human Oversight: High-impact AI decisions must not be left solely to machines; there must be meaningful human involvement with the authority to override automated outputs.
• Right to Appeal: India should establish dedicated tribunals for AI-related grievances, as current systems like RTI or consumer courts are ill-equipped to handle the technical nuances of algorithmic harm.
Judicial Scrutiny: All AI systems used in high-stakes public sectors, such as law enforcement and welfare, must be subject to constitutional scrutiny and Mandatory Algorithmic Impact Assessments before deployment. Conclusion India is at a constitutional crossroads where technological progress is rapidly reshaping the relationship between the State and the citizen. The shift toward algorithmic governance offers immense potential for efficiency and inclusion, yet it brings the quiet shadow of algorithmic injustice. Without a robust regulatory regime, automated systems risk institutionalizing historical biases and creating a "digital poorhouse" that punishes the marginalized through opaque and unchallengeable decisions. The importance of accountability cannot be overstated.
As the Supreme Court observed in Puttaswamy, efficiency is a facet of governance, but it must not be used to steam-roll fundamental freedoms. A true constitutional balance requires that every technological decision affecting rights is also treated as a legal decision, rooted in the values of equality, liberty, and dignity. Looking forward, India has the opportunity to bridge global regulatory models with its own constitutional ethos. By legislating with clarity and ensuring human-centric AI, the State can ensure that the digital revolution strengthens the Rule of Law and preserves the sanctity of the individual for generations to come
References • Constitution of India, 1950, Articles 14, 19, and 21. • Aadhaar Act, 2016. • Digital Personal Data Protection Act, 2023. • EU Artificial Intelligence Act (Regulation 2024/1689). Judicial Decisions • Justice K.S. Puttaswamy v. Union of India, (2017) 10 SCC 1. • Maneka Gandhi v. Union of India, (1978) 1 SCC 248. • Binoy Viswam v. Union of India, (2017) 7 SCC 59. • E.P. Royappa v. State of Tamil Nadu, (1974) 4 SCC 3. Government Reports & Academic Articles • NITI Aayog, Responsible AI for All, (2022). • Eubanks, V., Automating Inequality: How High-Tech Tools Profile, Police, and Punish the Poor, (2018). • Tiwari, S., Towards Rights-Based AI Framework in India, (2025).