On June 3, 2026, the Supreme Court released a preliminary draft of Regulations for Use of Artificial Intelligence in Courts, 2026 (“Draft Regulations“), inviting comments and suggestions from stakeholders and the general public until June 20, 2026. The Draft Regulations, prepared under the aegis of the Supreme Court’s Artificial Intelligence Committee, propose a comprehensive framework for the use and management of artificial intelligence (“AI“) in the Indian judicial system.
As stated in the accompanying notice, the Draft Regulations are grounded in the principles of human primacy, transparency, accountability, data protection, and judicial independence, and seek to establish a uniform institutional governance framework for responsible AI adoption across India’s judicial system.
The release of the Draft Regulations marks a significant and forward-looking step by the Supreme Court, recognising that AI is likely to play an increasingly important role in judicial administration and seeking to regulate its use through clear safeguards, rather than leaving its adoption unaddressed or open to interpretation.
Once notified, the Draft Regulations will apply to the Supreme Court, High Courts, District Courts, subordinate courts, tribunals and statutory commissions performing adjudicatory functions.
Human Oversight over AI-Assisted Judicial Decisions
The Draft Regulations place human decision-making and reasoning at the centre of the proposed framework. The relevant judicial officers have been given the leeway to use systems to assist in judicial and administrative functions, but they cannot use AI to replace the independent exercise of judicial authority. Thus, questions of law, fact, liability, rights and sentencing shall be determined by duly appointed judicial officers.
The Draft Regulations place the burden on the judicial officer for any decision taken with the assistance of AI. An officer cannot avoid accountability by relying on the opacity of an AI system, an inaccurate output or an AI “hallucination”. Further, the Draft Regulations state that AI-generated information is to be treated as advisory and must be verified before it is used. The Draft Regulations introduce the concept of the “Human-in-the-Loop” oversight, which is mandatory, under which AI outputs shall mandatorily be made subject to human review, supervision and verification, with final authority and accountability being on the human decision-maker.
Permissible Uses of AI under the Draft Regulations
The Draft Regulations contemplate a broad range of permitted uses, subject to prior approval, supervision and verification by the appropriate authority. These include case management, identification of defects in filings, preparation of cause lists, scheduling of hearings, transcription of court proceedings, translation of judgments and pleadings, legal research, precedent retrieval, citation verification and document summarisation.
The concerned judicial officers may use AI for assisting with administrative functions such as filings, record management, resource allocation, backlog monitoring and the automated preparation of notices and summons, in the prescribed formats. Further, under the Draft Regulations, AI tools may be used to assist litigants in understanding procedural requirements and for availing speech-to-text, text-to-speech, Braille translation and visual assistance. However, any transcription or translation generated using AI must be checked for accuracy.
The Draft Regulations also allow AI to be used for anonymising judgments, orders and other court records before they are published, for verifying the authenticity of documents or for identifying possible fraud. However, these uses remain subject to human review and no action may be taken solely on the basis of an AI-generated finding relating to fraud or document authenticity.
By keeping the permissible uses of AI broad, the Draft Regulations have sought to facilitate responsible adoption of AI, to create a system which may improve access to justice, reduce delays or enhance administrative efficiency. The list of permitted uses is illustrative and not exhaustive, allowing the appropriate authority to approve other uses of AI, provided that such use is consistent with the principles of human oversight, transparency and accountability.
Absolute Prohibitions under the Draft Regulations
The Draft Regulations identify certain uses of AI that are expressly prohibited. For instance, AI cannot be used by its human user to independently determine the final decision of the court, or be the sole basis of a judgment, decree or order. AI cannot be used to perform adjudicatory or sentencing functions, and any AI-generated output used by a judicial officer must be treated as advisory, subject to independent human evaluation.
Further, the use of AI for scoring any type of risk, such as flight risk, recidivism, bail eligibility or the credibility of a party or witness, is entirely prohibited. AI systems shall also not be used to predict, profile or infer the future behaviour of accused persons, litigants, witnesses or legal representatives. Similarly, AI-based surveillance or continuous monitoring of judges, advocates, litigants or other persons connected with court processes is prohibited, unless specifically authorised by law.
The Draft Regulations also protect the confidentiality of any deliberations carried out in courts, and any act that compromises such confidentiality is expressly prohibited. Collectively, these prohibitions recognise that although AI may assist courts in administrative and research functions, decisions affecting rights, liberty and credibility must remain transparent, accountable, fundamentally human and capable of being defended by the concerned judicial officer.
Data Protection and Sensitive Judicial Data
Data protection forms a central part of the Draft Regulations. All AI systems deployed in court processes must comply with the Digital Personal Data Protection Act, 2023, the Information Technology Act, 2000 and other applicable laws governing personal data and judicial information.
Any personal data which is processed using AI by a judicial officer must be carried out to the extent required to achieve the purpose of such processing and shall be kept confidential and private. The Draft Regulations create a distinct category of “sensitive judicial data”, which includes personally identifiable information concerning parties, witnesses and legal representatives, as well as information processed in connection with court proceedings, the unauthorised disclosure of which may cause harm.
Sensitive judicial data cannot be transferred to an external system without express written authorisation. Any permitted transfer must be supported by appropriate technical and contractual safeguards against unauthorised access, alteration, disclosure or misuse. Further, the Draft Regulations state that personal data must, to the extent technically feasible, be anonymised before being used to train, test or refine an AI system, subject to prior approval and compliance with applicable data protection law.
The Draft Regulations also limit access to personal data to a need-to-know basis and mandate annual review of the access protocols, with every AI system required to undergo periodic cybersecurity audits at intervals not exceeding one (1) year.
Disclosure of AI-Assisted Material
The Draft Regulations place disclosure obligations on the judicial officers and on the parties appearing before the court in cases where an AI tool has been used to materially assist in case management, document analysis or judicial administration which may affect the conduct of proceedings. Accordingly, under the Draft Regulations, the disclosure obligations apply both to judicial officers and to parties where an AI tool has been used for the preparation of the material placed before the court.
A party or legal representative using AI to prepare a document, pleading or evidentiary submission must disclose the AI-assisted nature of the material or any synthetic data or information used through a prescribed declaration or certificate. The Supreme Court may also require disclosure of the specific AI system used, the nature and extent of the assistance provided, and the measures taken by such a party to verify the accuracy of the output.
These disclosure requirements are to be enforced by duly allotting responsibility. The Draft Regulations expressly place the responsibility for fabricated, misleading or inaccurate material on the person submitting such information before the court, and any submission alluding to the fact that the content was generated by AI cannot be used as a defence.
The Apex Body and the Institutional Oversight Mechanism
The Draft Regulations propose a multi-tiered institutional framework led by a permanent Apex Body at the Supreme Court (“Apex Body“). The Apex Body will be responsible for setting the standards for AI use, developing adequate policies, approving acceptable AI systems and coordinating between the Supreme Court and High Courts, where separate AI committees will oversee implementation.
Every approved AI system would be required to be entered in an AI register and undergo technical, legal and ethical audits at least annually. Any AI-related errors, bias, security breaches and malfunctions would have to be recorded in an AI incident database and addressed through remedial measures. The Draft Regulations also require emergency fall-back protocols to ensure that essential judicial functions continue manually or through alternative systems if an AI tool fails or is suspended.
Conclusion
The Draft Regulations are the first step taken by the Supreme Court to establish a structured legal framework for AI use in the judiciary. While encouraging adoption of new technology and advanced tools, the Draft Regulations impose clear boundaries around adjudication, personal liberty, judicial independence and protection of sensitive judicial data. Under the Draft Regulations, courts, judicial officers, litigants, legal representatives and AI service providers will be required to adopt new processes related to approvals, disclosures, verifications, audits, data-protection and incident-reporting whenever AI is used in connection with court proceedings. However, the practical effectiveness of the Draft Regulations will ultimately be ascertained once it is implemented.
English
Chinese (Simplified)
Japanese