Supreme Court Sets AI Precedent Rulings Void, Mandates Zero-Tolerance

Supreme Court Annuls Judgments Built on AI Hallucinations

India's Supreme Court has declared void two judicial orders that relied on fabricated AI-generated case citations, establishing a binding zero-tolerance standard for the use of unverified artificial intelligence in legal proceedings. In Pooja Ramesh Singh v. Jammu and Kashmir Bank Ltd. [2026 INSC 668], a bench of Justice Pamidighantam Sri Narasimha and Justice Alok Aradhe annulled an NCLT order dated August 28, 2024, and an NCLAT judgment dated September 11, 2025—both of which had formed the basis for decisions in the underlying Essel Infraprojects insolvency dispute, according to India Today.

The court found that six AI-hallucinated judgments underpinned the annulled NCLT order—three judicial decisions that do not exist at all, and three others that were either misattributed to courts or cited out of contextual relevance. The ruling establishes that any quasi-judicial or judicial decision relying on fake, hallucinated, or non-existent precedents constitutes "no decision in the eyes of law," as documented by Drishti Judiciary.

Zero-Tolerance Doctrine and Professional Discipline

The Supreme Court's judgment goes beyond case annulment to establish institutional accountability. The bench directed the Bar Council of India to constitute an expert committee tasked with framing comprehensive guidelines and disciplinary consequences for advocates who cite unverified AI-generated precedents without verification. Citing such material is now formally classified as professional misconduct, per CaseMine.

The court emphasized the severity of the offense by invoking a comparison to industrial catastrophe. The bench likened the use of AI-generated fake precedents to "the release of methyl isocyanate"—a reference to the 1984 Bhopal disaster—to illustrate how the unchecked proliferation of false legal citations corrodes the rule of law itself, according to New Indian Express.

Procedural Remand and Broader Impact

The Essel Infraprojects insolvency matter has been remanded to the NCLT for fresh consideration on its merits, effectively nullifying months of prior adjudication. The Supreme Court held that courts and quasi-judicial bodies must adopt a "zero-tolerance mode" for producing, citing, or using AI-generated precedents without independent verification. This doctrine applies across all legal proceedings, not merely to this specific insolvency case.

The judgment signals that Indian courts will treat reliance on AI hallucinations as grounds for automatic annulment, even if the fabricated material represents only a marginal component of the decision-making process. The ruling arrives amid documented cases of legal AI tools producing fictional citations—a risk that has surfaced in jurisdictions worldwide—and establishes India as an early adopter of explicit, enforceable standards for generative AI use in judicial contexts.

The Bar Council's forthcoming expert committee will determine whether sanctions extend to suspension, fines, or other disciplinary measures against advocates who introduce unverified AI citations into court filings.