We use cookies to personalise the website and offer you the greatest added value. They are, among other purposes, used to analyse visitor usage in order to improve the website for you. By using this website, you agree to their use. Further information can be found in our data privacy statement.



India's Courts To Get a Rulebook for AI

​​​​​​​​​​​​​​​published on 22 June 2026 | reading time approx. 5 minutes​​​


After a prolonged regulatory vacuum, where India had no specific AI laws or guidelines, the Hon’ble Supreme Court of India has now unveiled a comprehensive draft framework to govern the use and deployment of  Regulations for Use of Artificial Intelligence in Courts, 2026​ (“Draft Regulations”), marking the first concrete step to regulate the use and integration of AI in courtrooms. 

By moving the discussion from abstract policy to concrete operational rules, the Draft Regulations establish clear standards for safety, transparency, and accountability in judicial technology. The draft is open for public comment until July 15, 2026. 

Introduction​

Most AI governance frameworks have, until now, been built around principles rather than rules. Fairness, transparency, and human oversight are values that almost everyone agrees on but agreeing on them and making them enforceable are two different things. The Draft Regulations attempt to bridge this gap by turning these principles into binding obligations, they establish: 
  1. A permanent oversight body at the Supreme Court level (Apex Body);
  2. AI Committees, and dedicated secretariats at every High Court;
  3. A research and evaluation centre to test AI tools before deployment (Centre of Research and Excellence on Artificial Intelligence (CoRE-AI)); 
  4. A public AI Register of approved tools and audit findings, and a live incident database to track errors, bias, and security failures over time.

The Draft Regulations are anchored in five core principles namely human primacy, transparency, accountability, data protection, and judicial independence that define the architecture of the framework. 

The scope of the draft framework is broad as it applies to all courts and tribunal in India, covering the use, deployment, and integration of AI across both judicial and administrative functions. It defines what AI may be used for, including legal research, document summarization, translation, transcription, and case management. It draws hard limits around prohibited uses such as outcome prediction, risk scoring, and surveillance. 

The Draft Regulations clearly state that the use of AI for judicial functions is permitted only in an assistive capacity. No judicial decision may be made solely on the basis of algorithmic output, and all AI-generated content is treated as advisory until verified by a human being.

Key Obligations​​

The Draft Regulations lay down a set of obligations governing the use of AI across court processes, directed at both legal professionals and AI technology vendors/ service providers.

Disclosure​​

Where AI-generated or AI-assisted content forms part of any document, pleading, or evidence filed before a court, the party or its representative must disclose this in a prescribed format. The disclosure obligation is not limited to content that is wholly AI-generated, it extends equally to material that has been drafted or refined with AI assistance.

Verification and recordkeeping​​

AI-generated output is treated as advisory in nature and cannot be placed before a court without prior human verification. The person relying on such output must maintain an auditable record of the generation process, including the AI system used, the source material relied upon, and the verification steps undertaken. 

Responsibility for errors​​

Where AI-generated content submitted before a court is found to be fabricated, hallucinated, or otherwise inaccurate, responsibility vests with the person who submitted it or who was responsible for its submission. Crucially, neither the conduct of the AI system nor the unintended nature of the error operates as a defence. This places a non-delegable duty of care squarely on the submitting party.

Contractual safeguards for vendors​​

Businesses that supply AI tools to courts must agree to a list of mandatory contract terms, including assignment of ownership or grant of a perpetual royalty-free licence to the court over tools developed using court data; restriction of sensitive judicial data to agreed purposes only; provision of technical and explainability documentation for high-risk tools; on-premises or sovereign cloud deployment for systems processing sensitive judicial data; audit rights in favour of the AI Secretariat; without-delay reporting of data breaches and AI-related incidents; and vendor indemnification of the court for harm arising from defects in the AI system.

Prohibited Uses​​

Separately, the rules draw a hard line around certain uses. These include outcome prediction, recidivism or risk scoring, individual profiling, surveillance, and the use of opaque or unexplainable AI systems in matters affecting personal liberty or legal rights. 

What it means for AI tool vendors and technology businesses​​

For businesses in the IT and ITES sector that develop or supply AI tools to courts, the Draft Regulations introduce a compliance environment that is both demanding and, in certain respects, still taking shape. The mandatory contractual terms, data residency requirements, audit obligations, and vendor indemnification provisions collectively represent a significant shift in how AI tools may be offered to courts as a customer. 

At the same time, some of the framework's key concepts remain undefined. The Draft Regulations do not specify what constitutes a "high-risk" AI tool, even though such tools attract heightened scrutiny and additional documentation obligations, leaving AI service providers uncertain about how to classify their products. The terms "undisclosed" and "opaque," used to describe prohibited AI systems, are similarly undefined, and the threshold for what counts as a sufficiently explainable system has not been spelled out.

There is also no clarity on the consequences where AI-generated content is found to be inaccurate or fabricated, and whether a court's response would be limited to procedural directions or whether more serious liability may follow. 

What to expect next​​

The Draft Regulations are still evolving, but the direction is clear as it marks a genuine shift in how AI governance is being approached in India, moving from stated principles to an enforceable institutional design. Once finalized, each court will need its own notification before the Draft Regulations take effect locally. Stakeholders of the judicial tech space may consider assessing their position, revisiting contractual arrangements where necessary, and engaging with the consultation process to seek greater clarity on liability standards and verification obligations.​​

Tech & Data Bites

author

Contact Person Picture

Prarthana Vasudevan

Consultant

+91 80 44784 803

Send inquiry

Rödl INDIA

Discover more about our offices in India​​. Re​​a​d more »​
Skip Ribbon Commands
Skip to main content
Deutschland Weltweit Search Menu