Code Cannot Replace Conscience: India's Draft Rules on AI in the Courts, 2026

Telegram Group Join Now

Relevance: General Studies Paper II — Polity & Judiciary; GS Paper III — Science & Technology, Artificial Intelligence; GS Paper IV — Ethics in AI

Source: Supreme Court of India, 2026

Code Cannot Replace Conscience: India’s Draft Rules on AI in the Courts, 2026

The Supreme Court’s AI Committee, chaired by Justice P. S. Narasimha, has unveiled the draft “Regulations for Use of Artificial Intelligence (AI) in Courts, 2026”, with public comments invited until June 20. The framework rests on one principle — AI must remain strictly subservient to human judgment, barring algorithms from the adjudicative sanctuary while harnessing them for administrative efficiency.

1 · Background and core concept

Adjudicative AI refers to algorithms that decide or recommend outcomes affecting personal liberty — bail, sentencing, credibility. Assistive AI, by contrast, is confined to administrative and accessibility functions.

The trigger: Tools like SUPACE (Supreme Court Portal for Assistance in Court Efficiency) and SUVAS (Supreme Court Vidhik Anuvaad Software, for translating judgments) are already in use — but lacked a regulatory framework.
Committee composition: Chaired by Justice P. S. Narasimha, with High Court judges Sanjeev Sachdeva, Raja Vijayaraghavan V, Anoop Chitkara and Suraj Govindaraj.
Statutory alignment: The draft mandates compliance with the Digital Personal Data Protection (DPDP) Act, 2023 — no personal data may be used to train or test judicial AI without prior approval.
Constitutional anchor: Judicial independence is part of the Basic Structure; final authority over law and liberty must vest exclusively in human officers.

2 · The bifurcation — what AI can and cannot do

Permissible · Admin Workflow optimisation Case management, cause-list preparation, hearing scheduling, defect scrutiny, anonymisation and record management.	Permissible · Access Language & accessibility Live transcription, regional-language translation, chatbots for litigants, and accessibility tools for persons with disabilities (PwDs).
Prohibited · Risk Scoring No predictive profiling No algorithmic assessment of bail eligibility, flight risk, recidivism, or witness credibility — these touch personal liberty.	Prohibited · Black-Box No opaque adjudication Absolute ban on unexplainable AI systems that could materially impact lawful rights or personal liberties of any citizen.

3 · Core analysis

A. The algorithmic bias problem

Mirror, not oracle: AI reflects its training data — historical prejudices on caste, gender or class get encoded and amplified.
Global warning: The US COMPAS risk-scoring tool was shown to mis-rate Black defendants as higher-risk — a cautionary tale the draft explicitly seeks to avoid.
Constitutional shield: The draft mandates non-discrimination on grounds of caste, religion, sex, language and economic status — anchored in Articles 14, 15 and 21.

B. The atrophy of judicial reasoning

Outsourced thinking: Heavy reliance on AI for precedent search and draft orders risks weakening the analytical muscle of clerks and junior judges.
Law as living craft: Legal reasoning requires reading societal shifts and occasionally breaking with precedent — a faculty algorithms cannot replicate.

C. Institutional safeguards

Proposed regulatory body: Two Supreme Court judges (one as ex-officio Chair), two High Court Chief Justices, two HC judges, a MeitY Joint Secretary, and independent experts in cybersecurity, finance and data privacy.
Data integrity: Strict DPDP Act compliance; no model training on personal data without prior approval.
Transparency mandate: Every AI output must be explainable and auditable — no black boxes.

4 · Way forward

Enforce Mandatory Human-in-the-Loop. Any AI output on adjudicative matters must remain advisory and non-binding, subject to independent judicial evaluation — never a substitute for the judge’s verdict.

Operationalise the regulatory body. Notify the proposed committee swiftly, with clear mandates on certification, audit and grievance redress for judicial AI tools.

Use AI to fight pendency. Deploy assistive AI on the 5.4 crore-case backlog via the e-Courts Mission Mode Project and National Judicial Data Grid (NJDG) — administrative speed, not adjudicative shortcut.

Build judicial AI literacy. Embed AI ethics modules in National Judicial Academy (NJA) training so judges can interrogate algorithmic outputs critically.

The draft 2026 regulations offer the world a workable template: algorithms for efficiency, humans for liberty. As India hosts the world’s most overburdened judiciary, technology must dismantle administrative bottlenecks — but the determination of guilt, innocence and human dignity must remain a deeply human endeavour, guided by conscience and equity.

UPSC Value Box

Draft AI in Courts Regulations, 2026	First Indian framework regulating judicial use of AI; public comments till June 20, 2026.
SC AI Committee	Chaired by Justice P. S. Narasimha; HC judges Sachdeva, Vijayaraghavan, Chitkara, Govindaraj as members.
SUPACE & SUVAS	Existing SC AI tools — SUPACE for case efficiency; SUVAS for translation of judgments into Indian languages.
DPDP Act, 2023	India’s data protection law; mandatory compliance for training judicial AI on personal data.
Basic Structure Doctrine	Judicial independence is part of the unamendable core of the Constitution (Kesavananda Bharati, 1973).
Article 14 & 21	Equality and Right to Life & Dignity — basis for banning discriminatory algorithmic adjudication.
Human-in-the-Loop	Global AI-ethics principle: a human must validate every consequential algorithmic decision.
National Judicial Academy (NJA)	Bhopal-based apex training institute for judges; proposed nodal trainer for judicial AI literacy.

Quick Revision

Draft “Regulations for Use of AI in Courts, 2026” — public comments till June 20, 2026.
SC AI Committee chaired by Justice P. S. Narasimha.
Permissible AI: case management, scheduling, transcription, translation, accessibility chatbots.
Prohibited AI: bail risk-scoring, recidivism prediction, witness credibility, opaque black-box adjudication.
SUPACE — Supreme Court Portal for Assistance in Court Efficiency; SUVAS — Supreme Court Vidhik Anuvaad Software.
DPDP Act, 2023 — compliance mandated for training judicial AI on personal data.
Basic Structure Doctrine — judicial independence held unamendable in Kesavananda Bharati (1973).
Proposed regulatory body includes 2 SC judges, 2 HC CJs, 2 HC judges, a MeitY Joint Secretary & independent experts.

Mains Practice Question

“Artificial Intelligence can ease the burden on India’s judiciary, but it cannot substitute the moral authority of a judge.” Critically examine, with reference to the draft Regulations for Use of AI in Courts, 2026. (15 marks · 250 words)

Structure hint:
Introduction — Anchor with the draft 2026 rules and Justice Narasimha-led committee.
Body Part 1 — Promise of AI: efficiency, accessibility, fighting pendency via SUPACE, SUVAS, NJDG.
Body Part 2 — Risks: algorithmic bias, atrophy of judicial reasoning, opaque black-box adjudication.
Body Part 3 — Safeguards: Basic Structure, Articles 14 & 21, DPDP Act compliance.
Way Forward — Human-in-the-Loop, regulatory body, NJA training in AI ethics.

Must mention:
Draft AI in Courts Regulations, 2026 ·
SUPACE & SUVAS ·
DPDP Act, 2023 ·
Basic Structure Doctrine ·
Human-in-the-Loop

Conclusion hint: Conclude that the draft offers a global template — algorithms for efficiency, humans for liberty — ensuring that India’s justice system remains guided by conscience, equity and constitutional values.

Start Yours at Ajmal IAS – with Mentorship StrategyDisciplineClarityResults that Drives Success

Your dream deserves this moment — begin it here.

Code Cannot Replace Conscience: India’s Draft Rules on AI in the Courts, 2026

Code Cannot Replace Conscience: India’s Draft Rules on AI in the Courts, 2026

1 · Background and core concept

2 · The bifurcation — what AI can and cannot do