The Supreme Court of India released a 35-page Draft Regulations for Use of Artificial Intelligence in Courts and invited public comments by mid-July, signaling a phased, non-mandatory rollout to govern AI use in the judiciary.

The draft outlines how AI can handle routine court tasks—such as transcription and translation, case management, and docketing—to speed up proceedings while keeping human judgment supreme.

It emphasizes core principles like access to justice, transparency, accountability, data protection, and preserving judicial independence, with human primacy at all decision points.

Private vendors may participate only with written court approval and must adhere to terms on data ownership, data sensitivity, and restrictions on retraining models with court data, with no exclusive IP rights over tools built on judicial data or public resources.

The framework prohibits high-risk uses such as predicting flight risk or recidivism, evaluating bail, or determining witness credibility, and forbids presenting AI outputs as independent evidence without disclosure.

Safeguards forbid training AI on personal data and prevent AI from making actual legal decisions like bail determinations; clarity is required when AI-generated content is used so that it is not mistaken for human judgment.

A lifecycle safety approach is mandated, including technical and ethical impact assessments, controlled testing, ongoing audits, an AI Register and Incident Database, and emergency fall-back protocols in courts.

Lithographic emphasis on transparency requires that litigants be informed when AI provides material assistance in their case, while other AI uses may remain undisclosed.

Chapter III details permissible AI use for administrative tasks such as transcription, translation, case management, scheduling, and anonymisation of judgments, with approval and supervision from designated AI committees.

An overarching governance structure includes an Apex Body at the Supreme Court, AI Committees at the SC and High Courts, an AI Secretariat, and a Centre of Research and Excellence on AI to set standards and monitor developments.

A data protection framework within the draft enforces purpose limitation, data minimisation, privacy by design, and safeguards for Sensitive Judicial Data, including regular security audits.

AI must remain subservient to human judgment, uphold natural justice, and prohibit discriminatory use, with strict rules on transparency and accountability.