Author: Anish Tandi
College: Centurion University of Technology and Management
Linkedin profile: | www.linkedin.com/in/anish-tandi-668a7b2a1
ABSTRACT
The Integration of AI into the Government decision-making processes in India marks a transformative shift in the public administration. From predictive policing and welfare benefit allocation to immigration screening and tax assessments, AI Systems are increasingly influencing decisions that affect fundamental rights. However, the Opacity of algorithms, lack of statutory oversight and potential for bias raise critical Constitutional questions. This Article examines the scope of judicial review over AI driven government actions within India’s existing Constitutional framework. It argues that while the principles of reasonableness, natural justice and non-arbitrariness under Article 14,19 and 21 remain applicable, traditional judicial review mechanisms face unique challenges when confronting algorithmic decision making. The Article analyses Judicial Precedents on the technology enabled governance, explores the doctrine of “Fairness accountability and transparency” in the algorithmic context and proposes a layered framework for reviewing AI based decisions. Ultimately, the Articles concludes that Indian Courts must evolve evidentiary standards, demand algorithmic explainability and assert jurisdiction over automated systems to preserve Constitutional accountability.
TO THE POINT
Artificial Intelligence is no longer a futuristic abstraction in Indian governance. The Central and State governments have deployed AI tools in diverse sectors, automated facial recognition systems (AFRS) for surveillance, the income Tax Department’s “Project Insight” for data mining, the Common Services Centres AI driven grievance redressal and algorithmic risk assessments in criminal justice. These systems generate outputs that directly affect citizens liberty, livelihood, and dignity.
The Core question is that, can Courts review such automated decisions, and if so, how? Under the Constitution, every administrative action affecting rights is subject to judicial review on grounds of illegality, irrationality and procedural impropriety. However, AI decisions challenge this framework because they lack identifiable human reasoning, operate as “Black Boxes” and evolve over time through machine learning.
Judicial review must therefore adapt. Courts must determine whether an AI decision satisfies the test of non-arbitrariness under Article 14, whether it violates the right to explanation implicit in Article 21 and whether automated decision making without human oversight amounts to excessive delegation. Without Judicial intervention, AI governance risks creating a “Rule of Algorithms” detached from constitutional morality.
USE OF LEGAL JARGON
The Constitutional framework for Judicial review flows from Articles 32 and 226 of the Constitutional of India. AI based government decisions constitute “State action” under the Article 12, as they are performed by or under the authority of the State. Consequently, such decisions must conform to the principles of audi alteram partem (hear the other side) and the duty to furnish reasoned orders, principles deeply embedded in administrative law.
Article 14 prohibits arbitrariness and mandates reasonable classification. However, algorithmic bais, whether based on race, caste, gender, or socioeconomic status, may constitute class legislation violative of equal protection. The Courts have employed the doctrine of proportionality to test whether governmental objectives justify rights infringing means. This doctrine becomes critical when AI Systems disproportionately impact vulnerable groups.
The principle of non-delegation, that essential legislative or adjudicative functions cannot be delegated to non-human entities requires re-examination. If an AI system effectively replaces judicial or administrative discretion, it may amount to abdication of function.
Further, the evidentiary standard of balance of probabilitiesand the burden of proof shift significant in AI cases. Without access to source code, training data or model architecture, citizens face an information asymmetry that renders the principles of equality of arms meaningless. Courts may need to invoke the doctrine of res ipsa loquitur (the thing speaks for itself) where opaque algorithms produce demonstrably harmful outcomes.
THE PROOF
The Empirical Evidence confirms the urgency of Judicial scrutiny. A 2023 report by the internet Freedom Foundation revealed that over 15 Indian government agencies deploy AI systems without any publicly available algorithmic impact assessment. The Telangana government’s use of AFRS in policing has been challenged before the Telangana High Court for alleged misidentification of protestors. The Delhi government’s AI based school admission lottery faced allegations of algorithmic glitches affecting thousands of students.
The Judicial pronouncements have already signed concern. In Justice K.S. Puttaswamy v. Union of India (2017), the Supreme Court recognized informational privacy and held that any technology interfering with privacy must satisfy the threefold test of legality, necessity and proportionality. Applying this to AI, any automated decision affecting privacy must be backed by a Valid law, not merely administrative convenience.
In Shreya Singhal v. Union of India (2015), the Court Struck down Section 66A of the IT Act for creating arbitrary standards. The Judgement’s emphasis on overbreadth and chilling effect applies equally to AI Censorship or content moderation tools that suppress speech without transparency.
International parallels offer cautionary tales. The Dutch “SyRI” case (2020) struck down an algorithmic welfare fraud detection system for violating human rights. The UK’s “R (Bridges) v. CCSWP” (2020) found the use if AFRS for policing disproportionate. India cannot afford to wait for comparable failures before asserting judicial oversight.
CASE LAWS
1. Justice K.S. Puttaswamy (Retd.) v. Union of India (2017) 10 SCC 1
The Supreme Court also held that the right to privacy is inherent in Article 21. Every act of the State (whether involving AI processing of data or not) must survive the test of legality, necessity and proportionality. This Judgement has been upheld as a Constitutional basis for attacking non-transparent algorithms.
2. Modern Dental College & Research Centre v. State of Madhya Pradesh (2016) 7 SCC 353
The Court drew the doctrine of proportionality under Indian Constitutional law. Decisions of AI which result in unnecessary of infringement of fundamental rights in relation to aims pursued, have to be annulled. Courts have to first consider less restrictive options for automated governance.
3. T.P. Senkumar v. Union of India (2017) 15 SCC 201
Despite considerations of human decision makers, the case also reaffirmed the principles that administrative decisions need to be reasoned, and open to review by the courts. Similarly, decisions made by AI which are not reasoned can be invalidated under Article 14.
4. K.S. Puttaswamy (Aadhaar) v. Union of India (2019) 1 SCC 1
The Court ruled that Authentication records in the scheme could only be retained for certain timeframes. This engenders transparency duties on AI Systems when dealing with Aadhaar Data. The Judgement also presupposed that auto-profiling does carry a Constitutional burden.
5. X v. Reserve Bank of India (2023) (pending, WP(C) No. 504/2023)
This petition is about unethical practices involved with AI algorithms used to make credit scoring and indebtedness decisions. Specifically, the petitioners contend that there is a violation of natural justice in cases where applications are refused credit automatically and without human intervention. This Outcome will greatly impact the course of AI governance jurisprudence.
CONCLUSION
Judicial Review of AI based government decision making is not an optional luxury but a Constitutional imperative. As India Embraces “AI for All” the risk of algorithmic arbitrariness grows proportionally. Courts cannot refuse Jurisdiction merely because technology is complex. Instead, they must develop now tools, appointment of technical commissioners, presumptions against opaque systems, mandatory algorithmic impact assessments and right to meaningful human review.
The principle of “bail is the rule, jail is the exception” teaches that liberty must be default. Similarly, when AI governs, transparency must be the rule and opacity the exception. Indian Courts have the doctrinal resources within Articles 14, 19 and 21 to meet this challenge. What remains is the will to apply Constitutional Scrutiny without being overawed by technological novelty.
FAQs
Q1. Can I challenge an AI generated government decision in Court?
Yes, any government action including AI driven decisions is subject to Judicial review under Articles 32 and 226 if it violates fundamental rights.
Q2. What legal ground can I use to challenge an AI decision?
You can argue that the decision is arbitrary (Article 14), violates natural justice, lacks transparency, or disproportionately impacts your rights under Article 21.
Q3. Does the government have to disclose how its AI works?
Courts have not yet ruled definitively, but principles of transparency and fairness strongly suggest that governments must disclose algorithmic logic affecting citizens’ rights.
Q4. Is there a Specific law regulating AI in Indian governance?
No, India currently lacks a dedicated AI regulation law. The Digital Personal Data Protection Act, 2023, provides limited safeguards for data processing but does not govern algorithmic decision making comprehensively.
REFERENCES
• Constitution of India, 1950, Articles 14,19,21,32,226
• Supreme Court of India. (2017). Justice K.S. Puttaswamyv. Union of India, 10 SCC 1.
• Supreme Court of India. (2019). K.S. Puttaswamy(Aadhaar) v. Union of India, 1 SCC1.
• Internet Freedom Foundation. (2023). State of AI Governance in India: An Empirical Study.
• District Court of the Hague. (2020). NJCM v. State of the Netherlands (SyRI Judgement) ECLI:NL:RBDHA:2020:1878.
• High Court of England and Wales. (2020). R (Bridges) v. Chief Constable of South Wales Police, EWCA Civ1058.
