Inside SC's proposed regulations for AI use in courts: What's allowed, what's absolutely barred

📌 Summary:

• The Supreme Court released draft Regulations for Use of Artificial Intelligence (AI) in Courts, 2026, to build a governance framework for AI in the judiciary; comments were sought from stakeholders and the public by July 15
• Phased rollout: they come into force for the SC on a date notified by the CJI, and separately for each High Court (and its subordinate courts/tribunals) on dates notified by that HC's Chief Justice; different provisions can start on different dates
• Permitted uses: courts must "actively seek" AI that improves access to justice, cuts delays or boosts efficiency — allowed for case management, transcription, translation, legal research, document summarisation, accessibility and administration, but only with written approval and officer supervision/verification
• Absolute, non-derogable bars: no judicial outcome may be reached by algorithmic decision-making alone; AI is barred for risk-scoring flight risk, predicting recidivism, evaluating bail eligibility, judging witness credibility, profiling future conduct, submitting AI output as independent evidence without disclosure, and using black-box (unexplainable) AI in matters affecting personal liberty
• Transparency and institutions: courts must inform parties when AI provides "material assistance"; architecture led by an SC-level "Apex Body" (SC/HC judges, a MeitY official, finance and cybersecurity experts) with five committees, HC-level AI Committees, an AI Secretariat, and a research body CoRE-AI

🎯 UPSC Relevance: GS2 (judiciary, judicial reform, e-governance, technology regulation, rule of law) with GS3 linkage (AI governance, ethics of automated decision-making).

📝 Prelims Facts:

• Draft: Regulations for Use of AI in Courts, 2026; public comments closed July 15
• Apex Body includes SC and HC judges, a MeitY official, and finance/cybersecurity experts; a research body CoRE-AI (Centre of Research and Excellence on AI) evaluates tools
• Human judicial authority remains determinative; AI use in decisions is only advisory

🔑 Key Term: Black-box AI — an AI system whose internal decision logic is not explainable/interpretable; barred in liberty-affecting matters.