Published on: 26th August 2026
Authored by: Sujal Tyagi
Vivekananda Institute of Professional Studies
Case Details
Case Name: Pooja Ramesh Singh v. Jammu and Kashmir Bank Ltd & Anr[1]
Citation: 2026 INSC 668; Civil Appeal No. 11950 of 2025
Court: Supreme Court of India (Civil Appellate Jurisdiction)
Bench: Justice Pamidighantam Sri Narasimha and Justice Alok Aradhe
Date of Judgment: 2 July 2026
Author: Sujal Tyagi
Provisions Considered: Section 7 of the Insolvency and Bankruptcy Code, 2016.[2]
Introduction
The decision in Pooja Ramesh Singh v. Jammu and Kashmir Bank Ltd & Anr represents a landmark Supreme Court ruling on the intersection of artificial intelligence, legal research, and judicial integrity. While arising out of an insolvency proceeding under the Insolvency and Bankruptcy Code, 2016 (IBC), the Supreme Court’s intervention established a strict standard regarding the use of AI-hallucinated or fabricated precedents by adjudicatory bodies and advocates alike.
Facts and Issues
Jammu and Kashmir Bank Ltd extended a credit facility of approximately ₹200 crore to Pan India Utilities Distribution Company Ltd (PIUDCL), a group entity. The facility was secured, among other things, by a corporate guarantee executed by Essel Infraprojects Ltd (EIL) and a mortgage over land located at Gorai, Borivali, Mumbai. Following PIUDCL’s default and non-performing asset (NPA) classification, the Bank filed an application under Section 7 of the IBC before the National Company Law Tribunal (NCLT), Mumbai Bench, seeking to initiate corporate insolvency resolution process (CIRP) against EIL as corporate guarantor for an outstanding debt of ₹87.43 crore.
Pooja Ramesh Singh, a suspended director of EIL, resisted the application. She contended that pursuant to a 2014 scheme of demerger and amalgamation sanctioned by the Bombay High Court, relevant assets and liabilities of EIL had been transferred to separate entities. She further argued that a subsequent sanction letter dated 18 November 2017 did not expressly continue the corporate guarantee, thereby releasing EIL from liability.
The NCLT rejected these arguments and admitted the Section 7 application on 28 August 2024, citing several precedents in its order. On appeal, the National Company Law Appellate Tribunal (NCLAT) affirmed the decision on 11 September 2025, relying on the same extracted precedents. The appellant subsequently filed Civil Appeal No. 11950 of 2025 before the Supreme Court.
Before the Supreme Court, counsel for the appellant demonstrated that the precedents cited by the NCLT and reproduced by the NCLAT could not be verified. Independent verification revealed that several cited judgments did not exist, while other citations referred to genuine decisions whose internal paragraphs had been replaced with fictitious text. An affidavit filed by the Bank confirmed that its counsel had not cited these fabricated cases; rather, the NCLT appeared to have sourced them during its own research process using an artificial intelligence tool.
Key Issues Before the Court:
1. Whether an adjudicatory order founded, even in part, on fabricated or AI-hallucinated precedents can be sustained under law.
2. Whether the legal consequence differs depending on whether such material was submitted by a litigating party or generated independently by the tribunal bench.
Arguments of the Parties
Appellant’s Submissions:
The appellant argued that the impugned orders were fundamentally vitiated because the reasoning of the NCLT, later affirmed by the NCLAT, relied materially on non-existent case law and fabricated textual extracts. Counsel submitted that this went beyond a standard legal error and struck at the core integrity of the adjudicatory process, rendering the orders void regardless of the commercial merits of the corporate guarantee dispute.
Respondent’s Submissions:
Jammu and Kashmir Bank Ltd did not defend the fabricated citations and confirmed via affidavit that its legal counsel had not relied upon them during oral arguments. The Bank’s submissions were restricted to the commercial merits, maintaining that EIL remained liable as a corporate guarantor notwithstanding the 2014 scheme and 2017 sanction letter, leaving the validity of the contaminated orders to the judgment of the Court.
Judgment and Ratio Decidendi
The Supreme Court set aside both the NCLT order dated 28 August 2024 and the NCLAT order dated 11 September 2025. It restored the Section 7 application to its original status before the NCLT for fresh disposal, preferably within two weeks, without being influenced by prior findings. The Court directed parties to maintain status quo in the interim and declined to express an opinion on the commercial merits of the guarantee dispute, focusing exclusively on decision-making integrity.
Ratio Decidendi:
1. Nullity of Contaminated Orders: A judicial or quasi-judicial decision founded on fake or hallucinated material is void in the eyes of the law and subverts the rule of law. Such an order must be set aside upon discovery, regardless of whether the fabricated authority was outcome-determinative.
2. Immateriality of Source: The origin of the fabrication does not alter the consequence. Whether a fake precedent is introduced by counsel or generated via independent bench research, the order remains equally unsustainable.
3. Professional Misconduct: Submitting unverified AI-generated precedents before a court constitutes professional misconduct by an advocate. The duty to verify citation authenticity before reliance is absolute, and ignorance of fabrication is no defense.
4. Human Control Over Adjudication: While AI tools may be used to enhance efficiency, human judgment must retain complete control over adjudication. Decision-makers must independently verify every citation at every stage.
5. Regulatory Directions: The Court directed the Bar Council of India to constitute a committee to evaluate the submission of fabricated AI-generated materials and establish binding regulatory guidelines and disciplinary rules for the legal profession.
Critical Analysis
Pooja Ramesh Singh is notable for creating a distinct doctrinal standard of judicial nullity. While Indian law has historically held that fraud vitiates judicial proceedings, this ruling establishes that an order contaminated by fabricated material is void even in the absence of intentional deceit by litigating parties. By extending this zero-tolerance standard to research conducted by the bench itself, the Court established a strict-liability framework for adjudicatory reasoning.
While aligned with international trends where courts have penalized advocates for submitting AI hallucinations, this decision goes further by addressing bench-sourced fabrications. However, the ruling presents practical challenges: it does not prescribe an immediate verification protocol (such as mandatory cross-referencing against authorized law reports like SCC Online prior to order execution), leaving forums with varying database access to interpret verification standards independently.
Additionally, in the context of insolvency law where expedited resolution is prioritized, setting aside orders in their entirety rather than severing tainted citations introduces procedural delays. Nevertheless, the decision emphasizes process integrity above speed, establishing a general precedent that human oversight must remain indispensable in automated legal research.
Conclusion
The ruling in Pooja Ramesh Singh v. Jammu and Kashmir Bank Ltd & Anr establishes a vital framework for legal practice in the digital age. By holding that AI-contaminated judicial orders are void regardless of source or impact, the Supreme Court has underscored that technological innovation must not compromise judicial rigor, professional ethics, or the rule of law.
References
[1] Pooja Ramesh Singh v. Jammu and Kashmir Bank Ltd & Anr, 2026 INSC 668.
[2] Insolvency and Bankruptcy Code, 2016, No. 31 of 2016, § 7 (India).
