Draft Regulations
The draft Regulations for Use of Artificial Intelligence in Courts, 2026 issued by the Hon’ble Supreme Court of India, adopt a pro-innovation approach while making it clear that Artificial Intelligence (“AI”) is an assistive tool and not a substitute for judicial decision-making.
The key takeaways are as follows:
1. Human Control Remains Paramount
The draft regulations are founded on the principle of “human primacy.” AI systems may assist judges and court personnel, but all decisions on law, facts and justice must remain with human judicial officers. AI outputs are advisory only and cannot replace judicial discretion or independent reasoning.
2. Permitted Uses of AI
AI may be used in courts for:
- Legal research and precedent retrieval;
- Citation verification;
- Summarization of pleadings and documents;
- Translation and transcription;
- Case management and scheduling;
- Litigant-facing chatbots and assistance tools;
- Accessibility services;
- Document authentication and fraud detection; and
- Anonymization of judgments and records.
The regulations therefore expressly encourage AI adoption for efficiency, access to justice and administrative improvement.
3. Absolute Prohibitions
The draft contains several non-derogable prohibitions:
- AI cannot determine judicial outcomes.
- AI cannot adjudicate disputes or impose sentences.
- AI cannot be used to assess bail eligibility, flight risk or recidivism.
- AI cannot evaluate witness credibility.
- AI cannot predict future behavior of litigants, accused persons or witnesses.
- AI cannot be used through opaque “black box” systems where rights or liberty are affected.
- AI cannot compromise judicial independence or confidentiality of deliberations.
These prohibitions cannot be relaxed even by the authorities created under the regulations.
4. Regulation of Generative AI
The draft specifically recognizes Large Language Models and Generative AI. Any AI-generated content proposed to be filed before a court must be disclosed and verified. The regulations expressly recognize the risk of AI “hallucinations,” including fabricated case law, legal principles, facts and evidence.
5. Disclosure Obligations on Lawyers and Parties
Where AI has been used in preparing pleadings, documents, evidence or submissions, the AI-assisted nature of the material must be disclosed to the court. Courts may require disclosure of:
- The AI system used;
- The extent of AI assistance; and
- The steps taken to verify accuracy.
If AI-generated material is false, fabricated or misleading, responsibility remains with the person who submitted it.
6. Explainability, Transparency and Accountability
The regulations require AI systems used by courts to be explainable, auditable and capable of human review. Accountability always rests with the responsible human officer. Neither an AI hallucination nor the opacity of a model can be used as a defense for an incorrect or harmful decision.
7. Data Protection and Judicial Confidentiality
Sensitive judicial data receives special protection. Court data cannot be freely used for model training and any use of personal data for training or refinement of AI systems requires approval and compliance with applicable data protection laws.
Conclusion
The draft regulations seek to create a “human-in-the-loop” AI framework. They encourage the use of AI for research, translation, accessibility and court administration, while prohibiting algorithmic adjudication and preserving human responsibility for all judicial decisions. For advocates, the most significant obligations are disclosure of AI-assisted work, independent verification of AI outputs and continuing responsibility for any inaccuracies generated by AI.
