- Introduction
Over 5 crore cases are pending before Indian Courts as of 2025, according to the National Judicial Data Grid. It is a number that has and also explains why, in April 2021, the Supreme Court of India quietly launched SUPACE. It was designed to assist judges in processing case-related information. The Supreme Court has described it as a tool that does not decide but assists. But the line between assistance and influence in adjudication is constitutionally significant. If an AI tool determines what legal material reaches a judge’s desk, it shapes the informational environment of justice. The article then states its thesis that SUPACE’s current deployment, absent any transparency or accountability mechanism, raises cognisable concerns under Article 21’s guarantee of procedural fairness, and that existing doctrine compels a regulatory response.
II. SUPACE -What It Is and What It Does
On April 6th 2021, then Chief Justice of India S.A. Bobde introduced the Supreme Court Portal for Assistance in Court’s Efficiency (SUPACE). As an artificial intelligence portal, SUPACE is designed to make relevant facts and laws available to a judge depending on the matter currently being heard. As a research tool for judges, its primary purpose is to ostensibly improve the Court’s efficiency and eventually begin to aid in reducing the pendency of cases.
The Supreme Court has consistently maintained a clear distinction: SUPACE assists, it does not adjudicate. It is a research tool, not a decision-maker. This distinction is legally significant and must be respected. This article does not argue that SUPACE is deciding cases or that judges are blindly following its outputs. Judges remain fully in command of their reasoning and their decisions.
However, as NITI Aayog’s Responsible AI for All (2021) acknowledges, even assistive AI systems exercise influence through the curation of information. The selection of which precedents to present, which facts to highlight, and which statutory provisions to foreground is itself a form of legal reasoning, one that is now being performed, at least in part, by an algorithm whose criteria are notpublicly disclosed. It is the foundation of the constitutional concern this article develops.
III. Article 21 and the Procedural Fairness Guarantee
Article 21 of the Constitution provides that no person shall be deprived of his life or personal liberty except according to procedure established by law. For the first three decades after the Constitution came into force, this provision was interpreted narrowly. In A.K. Gopalan v. State of Madras (1950), the Supreme Court held that “procedure established by law” meant simply any procedure prescribed by a validly enacted statute. The fairness of that procedure was not the court’s concern.
Maneka Gandhi v. Union of India (1978) changed everything. Justice P.N. Bhagwati, writing for a constitution bench, held that the procedure under Article 21 must be “right, just and fair not arbitrary, fanciful or oppressive.” The court went further and established what has since been called the golden triangle-Articles 14, 19 and 21 must be read together. Any law or procedure affecting personal liberty must satisfy the equality standard of Article 14, the freedom standard of Article 19, and the fairness standard of Article 21 simultaneously.
Embedded within this expanded Article 21 is the principle of audi alteram partem hear the other side. The Supreme Court in Olga Tellis v. Bombay Municipal Corporation (1985) confirmed that the right to a hearing is a component of the right to life and personal liberty under Article 21. A hearing, to be fair, requires that both parties have equal access to the information before the decision-maker. Neither party should be disadvantaged by information asymmetry that they could not anticipate or address.
This is precisely where SUPACE creates a constitutional tension. If a judge reviews AI-curated material before a hearing material that neither party has seen, whose basis of selection neither party knows, and which neither party can challenge the equality of the hearing is structurally compromised. One might argue that parties can still make their submissions freely. But submissions are made in response to a judicial frame of reference that has already been partially shaped by an algorithm. The parties are responding to a question they did not know had already been partially answered.
The European Court of Human Rights articulated the principle of equality of arms in Dombo Beheer B.V. v. Netherlands (1993) that each party must have a reasonable opportunity to present their case under conditions that do not place them at a substantial disadvantage vis-à-vis their opponent. While India is not bound by ECHR jurisprudence, the Supreme Court has drawn on it as persuasive authority in developing Article 21 most notably in Selvi v. State of Karnataka (2010), where the court used ECHR standards to interpret the right against self-incrimination under Article 20(3) read with Article 21. The equality of arms principle is therefore a legitimate doctrinal reference point in the Indian constitutional context.
IV. The Black Box Problem -Transparency, Reasons and Article 14
Beyond procedural fairness, SUPACE raises a distinct concern under the administrative law principle of reasoned decision-making and the arbitrariness standard of Article 14.
The Supreme Court in S.N. Mukherjee v. Union of India (1990) held that the duty to give reasons is a fundamental requirement of justice. It serves two purposes- it disciplines the decision-maker to think clearly, and it enables the affected party to understand the basis of the decision and to challenge it effectively. If AI is shaping the informational inputs of judicial reasoning and that influence is neither disclosed nor documented, the reasoning process acquires a layer of opacity that the duty to give reasons was designed to prevent.
This is what scholars have called the black box problem. As Frank Pasquale observes in The Black Box Society (2015), AI systems make consequential decisions through processes that are opaque even to their operators. The criteria by which SUPACE selects legal material which cases it prioritises, which it excludes, how it weighs competing precedents are not publicly disclosed. The algorithm’s internal logic is, in the language of computer science, a black box.
In the legal context, this opacity raises a cognisable Article 14 concern. In Shayara Bano v. Union of India (2017), Justice Nariman, writing for the majority, held that manifest arbitrariness action that is excessive, capricious or without adequate determining principle violates Article 14. An algorithmic system that curates judicial information without any disclosed, auditable, or challengeable criteria is arbitrary in precisely this sense. Neither the parties nor the public can verify that the system is operating consistently, impartially, or without embedded bias.
The data dimension adds a further layer. SUPACE processes the personal data of litigants, their case files, factual submissions, and legal strategies. The Digital Personal Data Protection Act, 2023 under Section 4 requires that personal data be processed only for specified, lawful purposes with appropriate safeguards. Section 8 imposes obligations on data fiduciaries to maintain data accuracy and implement security measures. The Supreme Court, as the entity deploying SUPACE, is a data fiduciary within the meaning of the Act. Whether SUPACE’s data processing architecture is fully compliant with the DPDPA has not been publicly examined.
This concern is reinforced by the Supreme Court’s own landmark ruling in K.S. Puttaswamy v. Union of India (2017), where a nine-judge bench unanimously held that the right to privacy including informational privacy and the right to control one’s personal data is a Fundamental Right under Article 21. Litigants whose case files are processed by SUPACE have a cognisable privacy interest in how that data is used, stored, and protected.
V. The Regulatory Vacuum and the Case for a Framework
The constitutional concerns identified above exist against the backdrop of a complete regulatory vacuum. No existing Indian law specifically governs the use of AI tools in the judiciary.
The Information Technology Act, 2000 regulates electronic records and intermediaries but contains no provision governing AI systems in courts. The DPDPA, 2023 addresses data protection broadly but creates no specific obligations for judicial AI. The Supreme Court Rules, 2013 govern court procedure but were framed before AI was a practical reality and are entirely silent on it. The NITI Aayog’s Responsible AI for All (2021) identifies principles of explainability, fairness and accountability but is a policy document with no legal force and no specific application to judicial AI.
Internationally, the EU AI Act (2024) represents the most comprehensive legal response to this gap. Under Article 6 read with Annex III, AI systems used in the administration of justice including tools that assist judicial authorities in researching and interpreting facts and law are classified as high-risk AI systems requiring mandatory transparency, independent conformity assessments, human oversight mechanisms, and technical documentation of their algorithmic logic. The EU’s classification is not merely regulatory prudence. It reflects a foundational value judgment that the administration of justice is an area where algorithmic opacity is constitutionally intolerable.
India must arrive at the same judgment through its own constitutional framework. Drawing from Maneka Gandhi, Puttaswamy, Shayara Bano, and the EU AI Act, the following minimum standards are constitutionally imperative:
Disclosure to parties– When SUPACE or any AI tool has been used in the processing of a case, the parties must be informed. This follows directly from Maneka Gandhi‘s requirement of a fair and transparent procedure and S.N. Mukherjee‘s duty to give reasons.
Algorithmic audit and explainability– SUPACE’s case selection and material presentation logic must be subject to periodic independent audit. The basis of its outputs must be documentable and explainable. This is demanded by Article 14’s arbitrariness standard as articulated in Shayara Bano.
Data governance compliance– Processing of litigants’ case data must be brought into explicit compliance with the DPDPA, 2023, with clear protocols on data retention, access, and security giving effect to the privacy right recognised in Puttaswamy.
Institutional oversight– A dedicated AI Judicial Oversight Committee, constituted under the Supreme Court’s administrative authority and comprising judicial officers, technical experts, and civil society representatives, should be established to monitor deployment, review complaints, and develop standards.
VI. Conclusion
SUPACE is a genuine and commendable effort to make the Indian judiciary more efficient. The vision animating it that technology can expand access to justice, reduce pendency, and support a judiciary burdened by over fifty million pending cases is constitutionally sound in its aspiration. Chief Justice Chandrachud’s articulation of technology as a vehicle for democratising justice deserves respect and support.
But constitutional aspirations require constitutional safeguards. The right to a fair hearing under Article 21 as expanded by Maneka Gandhi is not satisfied merely by the absence of a decision-making AI. It requires that the entire adjudicatory process, including its informational architecture, be transparent, accountable, and open to challenge. SUPACE in its current form, operating without disclosure obligations, audit requirements, or oversight mechanisms, does not fully meet this standard.
The Supreme Court observed in Puttaswamy that privacy “is not a gift of the Constitution it is a natural right which the Constitution recognises.” By the same logic, the right to a fair hearing does not become less fundamental because the tool that engages it is algorithmic rather than human. A court system that asks citizens to trust its fairness must itself be transparent about the processes through which that fairness is delivered. A regulatory framework is not an obstacle to judicial modernisation. It is its constitutional precondition.
THIS ARTICLE IS WRITTEN BY ANANYA YADAV FROM FACULTY OF LAW, BHU
REFERENCE :
1 NITI Aayog, Responsible AI for All, 2021, p. 34.
2 A.K. Gopalan v. State of Madras, AIR 1950 SC 27.
3 Maneka Gandhi v. Union of India, AIR 1978 SC 597, per Bhagwati J.
4 Olga Tellis v. Bombay Municipal Corporation, AIR 1986 SC 180.
5 Dombo Beheer B.V. v. Netherlands, Application No. 14448/88, ECHR, 27 October 1993.
6 Selvi v. State of Karnataka, (2010) 7 SCC 263.
7 S.N. Mukherjee v. Union of India, AIR 1990 SC 1984.
8 Frank Pasquale, The Black Box Society: The Secret Algorithms That Control Money and Information (Harvard University Press, 2015).
9 Shayara Bano v. Union of India, (2017) 9 SCC 1, per Nariman J.
10 Digital Personal Data Protection Act, 2023, Sections 4 and 8.
11 K.S. Puttaswamy v. Union of India, (2017) 10 SCC 1.
12 Information Technology Act, 2000.
13 Supreme Court Rules, 2013.
14 Regulation (EU) 2024/1689 of the European Parliament and of the Council (EU AI Act), Article 6 read with Annex III, para 8.
15 K.S. Puttaswamy v. Union of India, (2017) 10 SCC 1, per Chandrachud J (as he then was).
