Supreme Court Draft Rules for AI in Courts: No to Sentencing by Machines, Yes to Assistance Only

Jun 15, 2026

Overview

The Supreme Court’s AI committee has released draft regulations for using AI in courts. AI systems will be allowed only in an assistive role, not for judicial decisions, sentencing, or profiling witnesses. “Black box” AI that cannot be explained is banned. The draft is open for public comments until 20 June 2026.

First Ever AI Rules for Indian Courts

For the first time, the Supreme Court of India has written down rules for using artificial intelligence in courts. On 3 June 2026, a committee appointed by the Chief Justice made public the draft Regulations for Use of Artificial Intelligence in Courts, 2026 . The rules allow AI only to assist judges, not to decide cases or sentence accused persons. “Black box” AI systems that cannot explain how they reached a conclusion are completely banned. The draft is now open for comments from lawyers, litigants, and the general public.

Background: Why These Rules Were Needed

The Supreme Court has been watching the growing use of AI in lower courts with concern. In March 2026, a Bench headed by Justice P.S. Narasimha strongly criticised a trial court. That court had relied on fake judgments generated by an AI tool. The Bench said this was not just an “error in decision-making”. It called it judicial “misconduct”. That incident made it clear that India needed formal rules for AI use in courts. The draft regulations are the result of that concern.

Key Provisions of the Draft Regulations

The draft regulations draw a clear line between what AI can do and what it cannot do.

What AI Can Do (Permissible Uses)

AI can be used for tasks that help judges and lawyers. These include:

Legal research, finding past judgments, and checking citations.
Summarising long pleadings and judgments.
Translating judgments, orders, and legal documents.
Transcribing court proceedings automatically.
Managing case lists and scheduling hearings.
Providing chatbots to help litigants navigate court procedures.

What AI Cannot Do (Prohibited Uses)

The draft explicitly bans AI from the following:

Reaching judicial outcomes : AI cannot be used to decide the outcome of a case, find facts, or apply the law to facts.
Sentencing : No AI system can decide punishment without mandatory “human-in-the-loop” oversight. The final decision must be made by a judge.
Profiling parties or witnesses : AI cannot be used to predict the future conduct of accused persons, witnesses, or parties. It cannot assess flight risk, bail eligibility, or credibility of witnesses.
Opaque or unexplainable AI : Any AI system that works like a “black box” (meaning its internal reasoning cannot be explained) is banned if it affects the rights or liberty of a person.

Human-in-the-Loop: The Core Principle

The most important principle in the draft regulations is that AI must function solely in an assistive capacity . It must remain “strictly subservient to human judgment and judicial authority”. This means that every output from an AI system is only advisory. A judicial officer must independently evaluate it. No judicial outcome can be reached through algorithmic decision-making alone.

Why These Regulations Matter

The draft regulations recognise that AI can be a useful tool. It can save time and reduce the burden of routine work on judges. But it also recognises the dangers. AI systems can be biased. They can make mistakes. They can be “black boxes” that nobody fully understands. When a judge makes a mistake, that judge can be questioned and held accountable. When a black box AI makes a mistake, there is no clear accountability. The draft regulations ensure that human judges remain in charge.

Impact on Legal Education

The draft regulations are not just for judges. They also change what law students need to learn. Every law graduate will enter a courtroom where AI is present. They must know what AI can do and what it cannot. They must understand the ethical boundaries of using AI in legal drafting. They must be able to question whether an AI tool used by the prosecution or a litigant complies with the rules.

In August 2024, the Bar Council of India had already asked universities to include AI in their law curricula. The draft regulations give urgency to that directive. Law schools must now teach not just how to use AI tools, but also the professional responsibilities that come with them. A lawyer who submits an AI-generated petition with fake citations is not just being careless. According to the Supreme Court, that could be professional misconduct.

Public Consultation and Webinar

The Supreme Court AI committee has invited comments and suggestions from stakeholders and the general public on the draft regulations. The last date for comments is 20 June 2026 .

To discuss the impact of these draft rules, The Hindu is hosting a webinar titled “The AI syllabus in legal education” on 20 June 2026 at 5:00 p.m. The panellists include:

Krishna Deo Singh Chauhan , Associate Professor, Jindal Global Law School, who has written extensively on AI regulation and governance.
Sarthak Das , a final-year law student at Government Law College, Mumbai, who has published on AI and copyright law.
Professor Ashish Bharadwaj , Pro Vice Chancellor, WPU Goa, a widely published scholar on technology policy and innovation.

The webinar will be moderated by M. Kalyanaraman of The Hindu. The three best questions from the audience will receive a free online subscription to The Hindu.

A Human Touch: The Judge’s Concern

Why are judges worried about AI? Because they have seen what happens when AI goes wrong. In the March 2026 incident, a trial judge submitted a judgment that cited several Supreme Court decisions. Only problem: those decisions did not exist. An AI tool had invented them. The judge had simply copied the AI’s output without verifying it. The Supreme Court Bench said this was not a small mistake. It was a violation of judicial duty. The draft regulations are designed to prevent such incidents. They make it clear that AI is a helper, not a decision-maker.

What Happens Next

The draft regulations are now open for comments. After 20 June 2026, the committee will review all suggestions. The final regulations will then be notified. Once notified, they will apply to all courts in India – from the Supreme Court down to district courts and tribunals.

Conclusion

The Supreme Court’s draft regulations for AI in courts strike a careful balance. They allow AI to help judges with research, translation, and case management. But they keep the final decision-making power firmly in human hands. AI cannot sentence, cannot profile, and cannot act as a black box. These rules are a first for India. They set a direction for how technology and justice can work together without sacrificing fairness or accountability.

Exam-Focused Points

Draft regulations released on: 3 June 2026.
Official name: Regulations for Use of Artificial Intelligence in Courts, 2026.
Core principle: AI must function solely in an assistive capacity, strictly subservient to human judgment.
Permissible uses: Legal research, summarisation, translation, transcription, case management, chatbots.
Prohibited uses: Reaching judicial outcomes, sentencing, profiling parties/witnesses, assessing bail or credibility.
Black box AI: Opaque or unexplainable AI systems affecting rights or liberty are banned.
Human-in-the-loop: Mandatory for any sentencing or adjudicative function involving AI.
Triggering incident: March 2026, Justice P.S. Narasimha Bench chided a trial court for relying on non-existent AI-generated judgments, calling it judicial misconduct.
Public consultation deadline: 20 June 2026.
The Hindu webinar: 20 June 2026, 5:00 p.m., titled “The AI syllabus in legal education”.
Panellists: Krishna Deo Singh Chauhan (Jindal Global), Sarthak Das (GLC Mumbai), Prof. Ashish Bharadwaj (WPU Goa).
Impact on legal education: Law schools must now teach AI literacy, ethics, and professional responsibility in using AI tools.