When Law Meets Generative AI: Preserving Judicial Integrity through Human Verification

Abstract

Artificial intelligence has moved from being a peripheral technological tool to an increasingly consequential component of professional decision-making. The legal system presents a particularly sensitive setting because judicial reasoning depends upon the accurate identification, interpretation and application of authoritative sources. The Supreme Court of India's decision in Pooja Ramesh Singh v. Jammu and Kashmir Bank Ltd. therefore marks an important early intervention in the relationship between artificial intelligence and adjudication. The appeal arose from insolvency proceedings in which the National Company Law Tribunal, Mumbai (NCLT) relied upon several purported judicial precedents which were later found to be non-existent, incorrectly cited, or accompanied by paragraphs that did not appear in the cited judgments. The National Company Law Appellate Tribunal (NCLAT) subsequently affirmed the NCLT's conclusion without detecting the problem. The Supreme Court, on 2 July 2026, set aside both orders and directed a fresh consideration of the insolvency application. More significantly, the Court declared a zero-tolerance approach towards the production, citation or reliance upon AI-generated legal authorities without verification and directed the Bar Council of India to constitute a committee to formulate guiding principles and disciplinary consequences.