Supreme Court AI Regulations for Courts 2026: The Draft Explained

The Supreme Court AI regulations for courts 2026 are still a draft. On 3 June 2026 the Supreme Court published the draft “Regulations for Use of Artificial Intelligence (AI) in Courts, 2026” and invited comments until 20 June 2026. The draft allows AI only for assistive work such as case management, research, summarisation, translation and transcription, under human control, and prohibits AI-only judicial decisions, AI risk scoring for bail or recidivism, and undisclosed AI-generated evidence. As on 31 October 2026 we found no report of a final notified version.

Why the Supreme Court AI regulations for courts 2026 were proposed

The draft came at the end of a run of fake-citation incidents. Petitions in the Delhi High Court cited paragraphs that do not exist; the Bombay High Court quashed a tax assessment that relied on non-existent decisions; and in February 2026 the Supreme Court, in Gummadi Usha Rani v. Sure Mallikarjuna Rao, called a trial court’s reliance on fake judgments misconduct, not a mere error. Courts were also already using AI: SUVAS, a machine-assisted translation tool introduced in 2019, and SUPACE, a research assistance portal for judges launched on 6 April 2021. The Kerala High Court had issued its own policy for the district judiciary on 19 July 2025. A national framework was the next step.

What the draft permits and prohibits

Commentators summarising the Supreme Court AI regulations for courts 2026 describe the prohibitions as absolute. The common thread is “human primacy”: a judge may use AI to find and organise material, but the finding, the reasoning and the order remain human.

The governance structure the draft sets up

• An Apex Body at the Supreme Court: a permanent national body to set minimum standards, supported by specialist committees on technical, cyber security, infrastructure and finance, and data matters.
• An AI Committee in every High Court: to oversee governance and approve tools, with an AI Secretariat for operations, audits and incident reporting.
• An AI Register: a running record of every AI system used in courts, its approved scope, the provider, impact assessments and audits.
• Impact assessments and audits: technical and ethical assessment before deployment, regular audits, and testing in controlled environments first.
• Incidents: malfunctions, bias or errors to be reported to the AI Secretariat.
• Data and procurement: alignment with the Digital Personal Data Protection Act, 2023, and rules for engaging private vendors.
• Training and grievances: capacity building for judges and staff, and a grievance mechanism that preserves other legal remedies.

The draft runs to ten chapters, from preliminary definitions to miscellaneous provisions, and defines AI broadly as a machine-based system that generates decisions, predictions, recommendations or outputs.

What it means for advocates and litigants

Summaries of the draft Supreme Court AI regulations for courts 2026 report that AI-generated evidence should not be submitted without full and transparent disclosure, and that generative AI content used in proceedings must be disclosed and verified, with AI-assisted filings to carry a declaration. Until the final text is notified, these are proposals, not binding rules. But the direction is clear, and it matches what the Supreme Court has already decided in a binding judgment: in Pooja Ramesh Singh v. Jammu and Kashmir Bank Ltd. (2 July 2026) it held that citing AI-generated precedent without verification is misconduct by an advocate, and asked the Bar Council of India to frame norms.

Practical steps now:

1. Verify every authority and section yourself, whatever tool helped.
2. Keep a short note of where and how AI was used in each matter, so a disclosure requirement, if notified, costs you nothing.
3. Watch the Supreme Court website and your High Court’s notices for the final regulations and any High Court rules.
4. Never file AI-generated images, audio or documents as evidence without disclosing how they were made.

Frequently asked questions

Are the Supreme Court AI regulations for courts 2026 in force?

No, not as on 31 October 2026. The Supreme Court published them as a draft on 3 June 2026 with comments due by 20 June 2026, and they come into force only on dates notified by the Supreme Court and each High Court. We found no report of a final notified version.

Can judges use AI to write judgments in India?

Under the draft, no. AI-only decisions and AI adjudication without human control are prohibited, and research output is advisory, to be verified by the judge. Kerala’s July 2025 policy for its district judiciary already bars AI from arriving at findings, reliefs, orders or judgments.

Do lawyers have to disclose AI use under the draft regulations?

Summaries of the draft say AI-assisted filings and AI-generated evidence would need disclosure. Because the text is not final, check the notified version and any practice directions before relying on a particular disclosure format.

Is AI risk scoring for bail allowed?

No. The draft lists AI-based risk scoring for bail, recidivism or credibility among its prohibited uses, along with opaque AI systems wherever personal liberty or legal rights may be affected.

Quick quiz

Stay ahead of these rules with the AI for Lawyers and Law Students certificate course, ₹499, and for the wider policy picture, see AI Law and Governance in India, ₹599.

Sources

• Supreme Court of India, draft Regulations for Use of Artificial Intelligence (AI) in Courts, 2026, published 3 June 2026
• Pooja Ramesh Singh v. Jammu and Kashmir Bank Ltd., Supreme Court, 2 July 2026, 2026 INSC 668
• Gummadi Usha Rani v. Sure Mallikarjuna Rao, Supreme Court, order of 27 February 2026
• Kerala High Court, Policy Regarding Use of Artificial Intelligence Tools in District Judiciary, 19 July 2025
• Digital Personal Data Protection Act, 2023

This page is for education and general information, not legal advice for a specific case. Law as on 31 October 2026.