India Supreme Court: Pooja Ramesh Singh AI Zero-Tolerance Rule Voids Orders

• On July 2, 2026, the India Supreme Court established a zero-tolerance rule for judicial decisions based on AI-generated hallucinations.
• In Pooja Ramesh Singh v. Jammu & Kashmir Bank Ltd., the Court invalidated NCLT and NCLAT orders for relying on non-existent AI-generated case law.
• The NCLT's errors included quoting non-existent paragraphs, misattributing citations, inventing passages, and citing entirely fictitious judgments.
• The Supreme Court declared that any decision relying on such hallucinated material is legally void, emphasizing the critical need for judicial AI-generated case law verification.
• Despite AI's potential to transform legal processes, its probabilistic nature necessitates stringent human verification to prevent errors and uphold judicial integrity.

The India Supreme Court has unequivocally declared that any judicial decision relying on AI-generated hallucinated material is legally void, establishing a strict zero-tolerance policy.

In a landmark ruling issued on July 2, 2026, the India Supreme Court delivered its first substantive decision addressing artificial intelligence (AI) hallucinations in judicial proceedings. The Court, in the case of Pooja Ramesh Singh v. Jammu & Kashmir Bank Ltd., unequivocally set aside orders previously issued by both the National Company Law Tribunal (NCLT) and the National Company Law Appellate Tribunal (NCLAT). This decisive action was taken solely because the tribunals had relied on non-existent, AI-generated case law in their judgments.

This ruling established a strict zero-tolerance rule for any judicial decision found to be based on hallucinated AI-generated material. The Supreme Court explicitly held that such a decision is considered "no decision in the eyes of the law," effectively rendering it void. This precedent, centered around the Pooja Ramesh Singh AI zero-tolerance rule, underscores the critical importance of verifying all legal information, particularly when derived from AI tools, to maintain judicial integrity and the finality of legal outcomes.

The proceedings originated from credit facilities extended by Jammu & Kashmir Bank Ltd. to Pan India Utilities Distribution Company Ltd. (PIUDCL), which were secured by a corporate guarantee from Essel Infraprojects Ltd. (EIL). Following PIUDCL's default on its payment obligations, the Bank initiated an insolvency resolution process against EIL under Section 7 of the Insolvency and Bankruptcy Code, 2016 (IBC). On August 28, 2024, the NCLT admitted the insolvency application, basing its decision on six purported Supreme Court decisions.

Upon independent verification, the Supreme Court discovered that all six of these cited decisions were tainted by AI hallucinations. Two cases, including Canara Bank v. N.G. Subbaraya Setty, were valid but the NCLT had quoted non-existent paragraphs from them. Another judgment, Everest Kento Cylinders Ltd. v. Union of India, existed, but the citation provided pertained to a different case. Furthermore, one decision, SBI v. Shree Ram Urban Infrastructure Ltd., featured both an incorrect case name and an entirely fabricated passage. Most egregiously, three other decisions—ICICI Bank Ltd. v. Urban Infrastructure Real Estate Ltd., VS Dempo & Co. Ltd. v. Reliance Communications Ltd., and Sarbjit Singh v. Union Bank of India—were entirely fictitious, with their names and citations simply not existing. This extensive reliance on unverified generative AI output, rather than a mere clerical error, led to the NCLT NCLAT AI reliance overturned by the Supreme Court. The NCLAT further compounded the error by simply reproducing all six decisions without any verification during its appellate review, and both advocates and the appellate forum failed to scrutinize these fabricated references.

The Supreme Court's ruling highlights a fundamental tension within the legal system regarding the adoption of AI. As probabilistic tools, AI models operate by predicting the next word in a sequence based on statistical likelihoods, rather than by verifying facts. This inherent characteristic makes all AI tools susceptible to 'hallucinating'—generating false citations, quotes, or even entire case names. This vulnerability necessitates rigorous judicial AI-generated case law verification.

Despite these risks, the transformative potential of AI in legal workflows is undeniable. AI tools possess an exponential capacity to streamline court processes and enhance access to justice. The Supreme Court's own Draft Regulations for Use of AI in Courts, 2026, actively promote responsible AI as a catalyst for impactful innovation within the justice delivery system. These regulations specifically endorse legal research, precedent retrieval, citation verification, and document summarization as permissible applications of AI systems, provided they are used responsibly and with due diligence.

The India Supreme Court's decision firmly establishes that even the slightest presence of fake or hallucinated material generated by AI will automatically invalidate the underlying judicial decision. This robust stance underscores the Indian judiciary AI zero-tolerance policy, emphasizing that the integrity and finality of judgments cannot be compromised by unverified AI output. The ruling places a significant burden on all legal professionals and judicial bodies to implement stringent verification protocols for any information sourced from AI. This precedent serves as a critical reminder that while AI offers powerful tools for legal research and analysis, human oversight and verification remain indispensable to uphold the foundational principles of justice.

Lawyers and compliance officers must implement stringent verification protocols for all AI-generated legal research and citations, as the Indian Supreme Court has established a zero-tolerance rule, rendering decisions based on AI hallucinations invalid. This precedent creates a significant professional accountability risk for unverified AI use in legal proceedings.

Source: Original reporting via legal analysis.