SCAORA Urges Caution on Supreme Court’s Draft AI Regulations, Calls for Stronger Safeguards Against ‘Black Box’ Systems
The Supreme Court Advocates-on-Record Association (SCAORA) has submitted an extensive set of recommendations on the Supreme Court’s Draft Regulations for the Use of Artificial Intelligence (AI) in Courts, 2026, advocating a cautious and phased approach to AI adoption within the judicial system. The recommendations were presented to Justice P.S. Narasimha, who heads the Supreme Court’s Artificial Intelligence Committee, with the Association emphasizing that technological innovation must not come at the cost of judicial independence, transparency, or litigants’ rights.
The Supreme Court had released the draft regulations in June 2026 to establish a comprehensive governance framework for the responsible use of AI in courts. The proposed framework seeks to permit AI as an assistive tool while prohibiting its use in core judicial functions such as deciding cases, determining bail eligibility, assessing witness credibility, or replacing judicial decision-making. Following requests from stakeholders, the consultation period for public comments was extended until July 15, 2026.
SCAORA constituted a dedicated AI Sub-Committee, which undertook a clause-by-clause examination of the draft regulations before submitting its report. One of its principal recommendations is that AI should not be deployed in judicial functions directly affecting litigants unless robust legal and technical safeguards are first established. According to the Association, contemporary generative AI systems remain incapable of providing meaningful explanations for their outputs, making them unsuitable for high-risk judicial applications where accountability and transparency are indispensable.
A major concern raised by SCAORA relates to the continued use of opaque or “black box” AI systems. While the draft regulations require AI systems to be explainable, they also contemplate the deployment of opaque AI under heightened scrutiny. The Association argues that this creates an inherent contradiction because modern large language models and neural networks cannot fully explain the reasoning behind their outputs. It has therefore recommended an outright prohibition on black-box AI systems in high-risk judicial functions.
The report also highlights the problem of the “black box paradox.” Although the draft regulations place ultimate responsibility on judges and court officials using AI, SCAORA questions how meaningful accountability can be enforced when even developers cannot fully explain how complex AI models arrive at particular conclusions. The Association argues that assigning responsibility without ensuring explainability could expose judges and court administrators to avoidable institutional risks.
Another significant concern relates to AI hallucinations and automation bias. Referring to the Supreme Court’s recent zero-tolerance stance against AI-generated fictitious judicial precedents, SCAORA has recommended mandatory “human-in-the-loop” and “human-on-the-loop” safeguards for every AI-generated output used in court processes. It also cautioned that judges and registry officials working under heavy workloads may unconsciously place excessive reliance on machine-generated suggestions, thereby affecting independent judicial decision-making.
The Association has further urged greater safeguards for judicial data and digital sovereignty. It expressed reservations about sensitive court data being processed through AI infrastructure controlled by foreign technology companies and recommended that judicial data remain within sovereign Indian infrastructure. It also called for immediate audits of AI tools already deployed in the Supreme Court, including SUPACE, SUVAS, AI-assisted e-filing systems, real-time transcription services, and SuSahayak, to ensure compliance with privacy, security, and accountability standards.
SCAORA additionally questioned the constitutional foundation of the proposed regulations, observing that the draft does not clearly specify whether its authority flows from Article 145, Article 142, or merely serves as model guidelines. The Association argued that while Article 145 empowers the Supreme Court to regulate its own practice and procedure, it does not authorize the Court to impose binding AI regulations upon High Courts, which possess independent rule-making powers under the Constitution. It also recommended greater participation of the Bar in AI governance structures to ensure balanced oversight and practical implementation.
The recommendations are expected to play an important role as the Supreme Court’s AI Committee finalizes India’s first judicial AI governance framework. While broadly supporting the responsible use of artificial intelligence to improve efficiency, translation, transcription, legal research, and court administration, SCAORA has emphasized that the judiciary must adopt a precautionary approach, ensuring that AI remains a tool to assist judges rather than influence or replace judicial reasoning. The final regulations are likely to shape the future integration of AI across courts in India while balancing technological advancement with constitutional values, procedural fairness, and public confidence in the justice delivery system.
