The Supreme Court of India has, in a series of interventions through 2025-26, sounded an unprecedented alarm over the use of Artificial Intelligence (AI) hallucinations in judicial decision-making. In its most recent order, the Court set aside rulings of the National Company Law Tribunal (NCLT) and the National Company Law Appellate Tribunal (NCLAT) after discovering that the NCLT had relied on fictitious, AI-generated legal precedents in an insolvency case. The Bench went so far as to compare the danger of unchecked AI reliance in courts to methyl isocyanate — the poisonous gas responsible for the 1984 Bhopal gas tragedy — describing it as “invisible, insidious, and catastrophic by the time anyone notices.”
This is not an isolated observation. Earlier, on February 27, 2026, the same Bench of Justices P.S. Narasimha and Alok Aradhe had taken cognisance of a trial court relying on AI-fabricated case laws, terming it not merely an error but “judicial misconduct.” The Court has now directed the Bar Council of India (BCI) to constitute a dedicated committee to frame strict disciplinary norms for lawyers citing unverified AI-generated material, while a draft “Regulations for Use of Artificial Intelligence (AI) in Courts, 2026” has been floated for public consultation. This draft explicitly prohibits the use of AI in core judicial functions such as adjudication, sentencing, bail decisions, and assessment of witness credibility.
💡 Get AI-powered exam prep on your phone!
For UPSC and SSC aspirants, this development sits at the intersection of judicial administration, technology governance, and constitutional accountability — a combination increasingly tested in Mains answers on Governance (GS-II) and Science & Technology (GS-III), as well as Ethics (GS-IV) case studies on professional integrity.
Background and Context
The rapid, largely unregulated adoption of generative AI tools such as large language models in legal research has created a new category of risk: “hallucination,” where AI systems generate plausible-sounding but entirely fictitious case citations, statutes, or judicial quotations. Since neither lawyers nor, in some instances, judges verify these outputs against authentic databases, fabricated precedents have begun entering court records, threatening the foundational principle that judicial outcomes must rest on verified law and fact.
Five Important Key Points
- The Supreme Court set aside NCLT and NCLAT orders in an insolvency matter after finding that the NCLT had relied on AI-generated fictitious legal citations, holding this to be a serious lapse overlooked even by the appellate tribunal.
- On February 27, 2026, a Bench of Justices P.S. Narasimha and Alok Aradhe held that AI-fabricated case law relied upon by a trial court amounted to “judicial misconduct,” not a mere procedural error.
- The Court has directed the Bar Council of India to set up a dedicated committee to frame disciplinary norms and accountability mechanisms for lawyers who cite unverified AI-generated legal material.
- A draft “Regulations for Use of Artificial Intelligence (AI) in Courts, 2026” explicitly bars AI from performing core adjudicatory functions — including sentencing, bail eligibility, and witness credibility assessment — and is currently open for public consultation.
- The Court clarified that AI can remain a useful assistive tool for improving judicial efficiency but can never substitute independent human reasoning, judicial discretion, or professional accountability.
Constitutional and Legal Framework
The Indian judiciary derives its authority to regulate its own procedure and the conduct of officers of the court from Articles 129 and 142 of the Constitution, which empower the Supreme Court to punish for contempt and to pass orders necessary for complete justice. The Advocates Act, 1961, read with the Bar Council of India Rules, governs professional misconduct by lawyers, and it is under this framework that the BCI has been tasked with drafting new disciplinary provisions specific to AI misuse. Separately, the National Company Law Tribunal derives its adjudicatory powers under the Insolvency and Bankruptcy Code (IBC), 2016 — the very framework whose credibility is now threatened if tribunal orders are shown to rest on fabricated jurisprudence.
Governance Concerns: Erosion of Institutional Trust
The core danger identified by the Supreme Court is not merely technical error but the erosion of institutional trust. Judicial pronouncements operate on the presumption of verified reasoning; once fabricated case law infiltrates orders — even inadvertently — every subsequent citation and appeal built upon that order is compromised. The Court’s Bhopal gas analogy is deliberately stark: it signals that AI-hallucination risk is systemic, silent, and can propagate across appellate hierarchies before being detected, as happened when the NCLAT itself failed to catch the NCLT’s error.
Professional Accountability and the Draft AI Regulations
The 2026 draft regulations mark India’s first serious attempt at judicial AI governance. By excluding AI from adjudication, sentencing, bail decisions, and credibility assessment of parties or witnesses, the framework preserves the constitutionally mandated exercise of judicial mind under Article 21 (procedural fairness) and the principles of natural justice. The regulations also anticipate disciplinary liability for lawyers, marking a shift from AI being viewed as a neutral tool to being treated as a source of professional risk requiring due diligence obligations analogous to those in Rule 3 of the BCI’s Standards of Professional Conduct.
Comparative Global Context
India’s response follows global precedent: courts in the United States, the United Kingdom, and Canada have similarly sanctioned lawyers for filing AI-fabricated citations since 2023, with some facing monetary penalties and referrals to bar disciplinary bodies. India’s approach is comparatively more structural, given that a fabricated citation entered not through a private practitioner’s oversight but through the reasoning of a specialised tribunal itself, revealing a gap even at the institutional level, not merely the individual lawyer level.
Implementation Challenges
Enforcing AI accountability faces several practical hurdles: the absence of mandatory AI-output verification protocols in tribunals; the uneven digital literacy of judicial officers regarding AI limitations; the difficulty of tracing which specific tool generated a fabricated citation; and the risk that overly restrictive regulation could deter legitimate AI-assisted legal research that genuinely improves efficiency in a judiciary burdened by pendency exceeding five crore cases nationally.
Way Forward
A multi-pronged approach is needed: mandatory citation-verification software integrated into tribunal case-management systems; compulsory AI-literacy modules in judicial academies and law schools; a graded disciplinary framework distinguishing inadvertent reliance from wilful fabrication; and a national database of verified precedents accessible to tribunals to reduce dependence on unverified AI searches. The BCI committee’s terms of reference should also extend to certifying AI legal-research tools before they can be relied upon in submissions.
Relevance for UPSC and SSC Examinations
GS-II (Governance, Constitution, Judiciary): Judicial accountability, tribunal reforms, Bar Council regulatory powers, Articles 129 and 142. GS-III (Science and Technology): AI governance, ethical use of emerging technology in public institutions. GS-IV (Ethics): Professional integrity, accountability in the use of technology, case study application. SSC Relevance: Constitutional bodies (Bar Council of India), landmark Supreme Court observations, IBC framework, NCLT/NCLAT structure. Key Terms: AI hallucination, NCLT, NCLAT, Insolvency and Bankruptcy Code, Bar Council of India, judicial misconduct, Articles 129 & 142.