Draft SC rules prohibit use of AI for judicial outcomes, witness profiling

Draft SC Rules Prohibit Use of AI for Judicial Outcomes, Witness Profiling

Regulations for Use of Artificial Intelligence (AI) in Courts, 2026 — UPSC Study Note

1. At a Glance

• The Supreme Court of India's AI Committee released a preliminary draft of the 'Regulations for Use of Artificial Intelligence (AI) in Courts, 2026' on June 3–4, 2026, open for public comment until June 20, 2026. [S1][S2]
• The draft establishes a foundational principle: AI must remain "strictly subservient to human judgment and judicial authority" and function solely in an assistive capacity. [S3]
• The regulations explicitly prohibit AI use in determining judicial outcomes, sentencing, bail, witness/party profiling, predictive scoring, and deployment of "opaque or unexplainable" AI systems. [S1][S2]
• Critical for UPSC: sits at the intersection of GS-II (Judiciary, Governance), GS-III (Technology), and GS-IV (Ethics in emerging technology); likely to generate Mains and Essay questions.

2. Why in the News

• The Supreme Court AI Committee made the draft public on June 4, 2026, inviting stakeholder comments and public feedback till June 20, 2026. [S1][S2]
• The draft follows a March 2026 rebuke by a Bench headed by Justice P.S. Narasimha, which chided a trial court for citing non-existent AI-generated judgments ("hallucinations"), calling it not just an error but judicial misconduct. [S3]
• In December 2025, the Supreme Court reconstituted its AI Committee under CJI Surya Kant to oversee deployment of AI tools in courts. [S4]
• A December 2025 petition before a Bench led by CJI Surya Kant and Justice Joymalya Bagchi had demanded regulation of AI and ML in the judicial system, with the CJI assuring that AI would "never overpower judicial decision-making." [S5]

3. Background & Evolution

• Predecessor initiative: SUPACE (Supreme Court Portal for Assistance in Court Efficiency) — an AI research tool already in limited use for legal research assistance.
• The regulations are framed against the background of the Digital India mission, National e-Governance Plan, and the e-Courts Project (Phase III).

4. Core Static Facts

Nomenclature - Full title: 'Regulations for Use of Artificial Intelligence (AI) in Courts, 2026' - Status: Preliminary/Draft (as of June 2026); not yet enacted. - Drafting body: Supreme Court AI Committee

Committee Composition [S1][S2] - Chair: Justice P.S. Narasimha (Supreme Court Judge) - Members: Justice Sanjeev Sachdeva, Justice Raja Vijayaraghavan V. - Apex Body (proposed): Chaired by a SC Judge nominated by the Chief Justice of India (CJI); includes High Court Chief Justices/Judges, a representative of MeitY (Ministry of Electronics & IT), and experts in finance, cybersecurity, and technology law.

Governance Structure [S2] - Apex Body at the Supreme Court level (permanent, full-time). - Oversight cascades to High Courts and subordinate courts.

Key Regulation - Regulation 43: Lawyers and litigants must disclose AI-assisted filings; courts may seek details of the AI system used. [S2]

Permitted Uses of AI (draft allows) [S1][S2] - Legal research and citation verification - Summarising pleadings and judgments - Translation and transcription - Drafting assistance - Case listing, scheduling, case management - Litigant-facing chatbots

Prohibited Uses of AI (draft bars) [S1][S2][S3] - Determining judicial outcomes / rendering judgments - AI-assisted sentencing without mandatory human oversight - Bail determinations / risk scoring - Profiling parties or witnesses (predictive profiling) - Surveillance of judicial officers - Use of "opaque" or "unexplainable" AI systems in any court process

Public Consultation Deadline: June 20, 2026 [S3]

5. Multi-Dimensional Analysis

Legal / Constitutional

• The regulations derive legitimacy from the SC's inherent powers under Article 142 and its rule-making authority under the Constitution of India.
• Directly addresses Article 21 (right to fair trial / life and liberty): AI profiling of witnesses or bail risk scoring could infringe due process and personal liberty.
• The bar on "opaque" AI aligns with the right to a reasoned order, a cornerstone of natural justice (audi alteram partem). [S2]
• The draft creates a new disclosure obligation (Reg. 43) — analogous to conflict-of-interest norms but for algorithmic tools. [S2]

Ethical / Governance

• Positions human judicial authority as non-delegable; AI remains a tool, not a decision-maker — reflects the principle of accountability of public office.
• Prohibition on AI profiling addresses algorithmic bias: AI trained on historical data could entrench caste, class, or gender biases in sentencing or bail decisions.
• Opacity/explainability requirement mirrors EU's AI Act (2024) high-risk AI obligations — sets India on a comparable trajectory.
• Disclosure norm (Reg. 43) advances transparency, counters "AI-washing" in legal arguments. [S2]

Scientific / Technological

• Distinguishes assistive AI (permissible: NLP for research, translation, summarisation) from decisional AI (prohibited: sentencing algorithms, risk scorers).
• The March 2026 hallucination incident highlights the reliability problem of Large Language Models (LLMs) in high-stakes settings. [S3]
• Requirement for explainable AI (XAI) excludes black-box deep-learning models from court use — aligns with global best practice.
• MeitY representation on the Apex Body ensures technical expertise in governance. [S2]

Administrative

• Apex Body model creates a centralised regulatory hub at the SC with downward cascade to HCs — addresses consistency across 25 High Courts and ~19,000 subordinate courts.
• Risk: varying digital infrastructure across courts could make uniform compliance difficult.
• The draft's public consultation mechanism reflects participatory rulemaking — unusual for judicial rule-making in India.

Social

• Bars witness profiling — protects vulnerable witnesses (women, minorities, marginalised communities) from algorithmic stereotyping.
• Prevents risk-scoring for bail, which disproportionately disadvantages under-trial prisoners from disadvantaged backgrounds (India has ~75–77% under-trials in prison population per NCRB data).
• Chatbot assistance for litigants could improve access to justice in regional languages if properly implemented.

Historical

• Mirrors judicial caution seen globally: US COMPAS algorithm controversy (Loomis v. Wisconsin, 2016) — AI risk-scoring in sentencing challenged on due process grounds.
• EU and UK courts have similarly debated boundaries of AI in judicial processes; India's draft is among the first comprehensive court-specific AI regulatory frameworks in Asia. [S1]

6. Recent Developments (Last 12–18 Months)

• December 2025: CJI Surya Kant reconstitutes SC AI Committee; petition heard seeking AI/ML regulation in courts; CJI asserts AI will never overpower judicial decision-making. [S4][S5]
• March 2026: Justice P.S. Narasimha bench censures trial court for citing non-existent AI-generated ("hallucinated") judgments; labels it judicial misconduct. [S3]
• June 3–4, 2026: SC AI Committee releases preliminary draft 'Regulations for Use of AI in Courts, 2026'; public comments invited until June 20, 2026. [S1][S2][S3]

7. Prelims Hooks

1. The draft 'Regulations for Use of Artificial Intelligence (AI) in Courts, 2026' was released for public consultation on June 4, 2026. [S1]
2. The SC AI Committee drafting these regulations is chaired by Justice P.S. Narasimha. [S3]
3. Public comment deadline on the draft regulations: June 20, 2026. [S3]
4. Regulation 43 of the draft mandates disclosure of AI-assisted filings by lawyers and litigants. [S2]
5. The draft bars AI from sentencing, bail determination, risk scoring, and witness/party profiling. [S2]
6. AI systems used in courts must be "strictly subservient to human judgment and judicial authority" per the draft. [S3]
7. The proposed Apex Body is to be chaired by a Supreme Court Judge nominated by the Chief Justice of India. [S2]
8. MeitY (Ministry of Electronics and Information Technology) has a representative on the proposed Apex Body. [S2]
9. The draft prohibits "opaque" or "unexplainable" AI systems from any court process — mandating explainable AI (XAI). [S1]
10. In March 2026, a bench chided a trial court for relying on AI-generated non-existent judgments — termed judicial misconduct, not merely a decision-making error. [S3]
11. Permitted AI uses include: legal research, citation verification, summarisation, translation, transcription, drafting assistance, scheduling, and litigant chatbots. [S2]
12. The SC AI Committee was reconstituted by CJI Surya Kant in December 2025. [S4]
13. AI-assisted surveillance of judicial officers is explicitly prohibited under the draft. [S2]
14. The regulatory framework governs all courts — SC, High Courts, and subordinate courts across India. [S2]

8. Mains Relevance

Plausible Mains Questions: 1. "The Supreme Court's draft AI regulations for courts reflect a cautious but necessary approach to technology in the justice system. Critically examine their significance for judicial independence and access to justice." (GS-II/IV) 2. "Algorithmic bias in judicial processes poses a threat to constitutional guarantees of equality and liberty. In the context of India's draft court AI regulations, discuss the challenges of ensuring fair and explainable AI in the justice system." (GS-II/III/IV) 3. "Should AI be allowed in judicial decision-making? Examine the ethical, constitutional, and practical dimensions of regulating artificial intelligence in Indian courts." (GS-IV/Essay)

9. Related Topics to Study Next

10. Common Errors / Trap Areas

1. Wrong committee chair: Aspirants may confuse Justice P.S. Narasimha (committee chair) with CJI Surya Kant (who reconstituted the committee). These are distinct roles.
2. Draft ≠ enacted law: As of June 2026, these are draft/preliminary regulations, not binding rules. Do not state them as in force.
3. Permitted vs. Prohibited mix-up: AI for scheduling and case listing is permitted; AI for sentencing and bail is prohibited. Aspirants frequently invert this.
4. MeitY confusion: The relevant ministry on the Apex Body is MeitY (Electronics & IT), not the Ministry of Law & Justice — a common substitution error.
5. Scope confusion: These regulations govern courts (judicial institutions), not the broader AI regulatory framework for the economy (which falls under MeitY's draft National AI Policy). They are distinct instruments.

11. Sources

• [S1] India's Supreme Court Proposes Comprehensive Framework To Govern Artificial Intelligence In Courts — https://www.mondaq.com/india/new-technology/1798568/indias-supreme-court-proposes-comprehensive-framework-to-govern-artificial-intelligence-in-courts — (Tier 4 / legal commentary)
• [S2] Supreme Court Releases Draft AI Regulations For Courts; Judges To Retain Final Authority — https://knnindia.co.in/news/newsdetails/sectors/supreme-court-releases-draft-ai-regulations-for-courts-judges-to-retain-final-authority — (Tier 4)
• [S3] Draft SC rules prohibit use of AI for judicial outcomes, witness profiling — The Hindu, June 5, 2026 — https://www.thehindu.com/todays-paper/2026-06-05/th_international/articleGL9G2QG4R-14835364.ece — (Tier 4 — article excerpt provided as primary source)
• [S4] Supreme Court Reconstitutes AI Committee to Oversee Adoption and Deployment of AI Tools — https://www.newsonair.gov.in/supreme-court-reconstitutes-ai-committee-to-oversee-adoption-and-deployment-of-ai-tools — (December 2025) — (Tier 1 / government news service)
• [S5] Supreme Court Assures Judges Exercising Utmost Caution on Use of AI in Judicial Process — https://www.newsonair.gov.in/supreme-court-assures-judges-exercisingutmost-caution-on-use-of-ai-in-judicial-process/ — (December 2025) — (Tier 1 / government news service)