4 min readJul 22, 2026 06:45 AM IST First published on: Jul 22, 2026 at 06:45 AM ISTAlso by Pavithra ManivannanArtificial Intelligence is coming to India’s courts, not as a distant possibility but as an active policy question. This June, the Supreme Court published draft regulations for the use of AI in courts for public comments. They govern how AI may be used across the entire judiciary, create institutions to oversee their adoption and prescribe approval processes for AI systems.AdvertisementBut courts are only beginning to use AI, and no one fully understands yet how it will behave in practice. These regulations will need to evolve as experience accumulates.The more important question, then, is whether the foundations are right. Two things in particular are hard to reverse once set, and will shape the integrity of everything that follows. Both need to be answered before AI enters the courtroom.The first is about institutions. Every wave of technology use in India’s judiciary has created a new governance body. Computerisation brought computer committees. AI is now producing an apex body, five standing committees, High Court AI committees, district AI secretariats, and a new Centre of Research and Excellence.AdvertisementThe problem is architecture. Institutional structures must not be built around individual technologies. Computerisation, digitisation, video conferencing, and AI are successive tools in the same transformation of court administration. For instance, an AI-assisted scheduling system is simultaneously a case management, infrastructure, digitisation and AI project. If each of these tools and functions has separate committees, it fragments accountability and slows decision-making. The judiciary needs a permanent institution capable of absorbing the functions currently held by the e-Committee alongside the new AI governance mandate, and whatever technology comes next. India has done this before: TNMSC professionalised medicine procurement in Tamil Nadu and became a model for other states; UIDAI built and maintains the Aadhaar ecosystem. Execution requires permanence, expertise and clear ownership — none of which committees with rotating members can provide.The second is about standards. When India built its payments infrastructure, it made one decision that changed everything: It separated the standard from the market. NPCI wrote the specification for UPI and stepped back. It didn’t build PhonePe or Google Pay — the market did. The standard was shared; the innovation happened around it. India’s courts are now at a similar inflection point with AI.The Draft Regulations establish who approves AI systems for court use and how. What they do not establish is what any system must actually demonstrate to earn that approval. This is not a minor gap. Before a pharmaceutical company can sell a drug in India, its manufacturing process must meet internationally defined quality standards.you may likeFor judicial AI, the equivalent is missing. What must a system explain, and to whom? What error rate is acceptable in a bail matter versus a scheduling task? Where does training data come from? Without these answers defined in advance, each of India’s 25 high courts must independently determine what constitutes an acceptable AI system. Given the wide variation in technical capacity across the judiciary, they will arrive at different answers. The result is inconsistent approvals — and inconsistent justice.Standards cannot be set by the body that procures against them. The Draft Regulations already envision a Centre of Research and Excellence on AI (CoRE-AI). It should be set up as a neutral body with no stake in which vendor wins, no procurement budget to defend. A permanent technical institution should procure against them. A judicial Committee should approve them. Just as NPCI stepped back and let the market build, CoRE-AI’s value lies precisely in staying out of the game it is refereeing. The SC must push for foundations that will hold not just for today’s AI, but for whatever comes next.Sankaran is a co-founder of PUCAR and Agami. Manivannan is research lead at XKDR Forum