Authored By: Sakshi Manisha Nitin Parulekar
Siddharth College Of Law (University of Mumbai)
- CASE CITATION AND BASIC INFORMATION
Case: Pooja Ramesh Singh v Jammu and Kashmir Bank Ltd. & Anr.
Citation: 2026 INSC 668
Case No.: Civil Appeal No. 11950 of 2025
Court: Supreme Court of India
Date of Decision: 2 July 2026
Bench: Justice Pamidighantam Sri Narasimha and Justice Alok Aradhe
Area of Law: Artificial Intelligence and judicial process; professional ethics; insolvency law; administration of justice. (Supreme Court of India)
- INTRODUCTION
Artificial Intelligence (AI) has rapidly transformed professional and intellectual activities, including legal research, drafting, document review and legal education. Generative AI tools can produce legal arguments and identify apparently relevant judicial decisions within seconds. At the same time, these systems may produce inaccurate or completely fictional information, commonly referred to as “AI hallucinations.” The legal profession faces a particularly serious problem when such fabricated material is presented as genuine judicial precedent.
The Supreme Court of India addressed this emerging problem in Pooja Ramesh Singh v Jammu and Kashmir Bank Ltd. & Anr., decided on 2 July 2026. The case is significant because the National Company Law Tribunal (NCLT) had relied upon several judgments that were either non-existent, incorrectly cited, or contained paragraphs that could not be found in the judgments cited. The Supreme Court found that the material appeared to have been generated through AI and held that a decision founded upon fake or hallucinated precedent is “no decision in the eyes of the law.” (Indian Kanoon)
The judgment is particularly contemporary because it does not reject AI as such. Instead, it recognises AI as a potentially useful aid to legal work while insisting that human beings must retain control over judicial reasoning and verification. The Court expressly distinguished between the legitimate use of AI and the irresponsible reliance upon unverified AI-generated legal authorities.[1]
- FACTS OF THE CASE
The appellant, Pooja Ramesh Singh, was a suspended director of Essel Infraprojects Ltd. (EIL), which was described as the corporate debtor and corporate guarantor of Pan India Utilities Distribution Company Ltd. (PIUDCL). PIUDCL had obtained loan facilities from Jammu and Kashmir Bank Ltd. To secure those facilities, EIL had executed a corporate guarantee in favour of the Bank. Subsequently, PIUDCL experienced financial difficulties and defaulted on its repayment obligations, resulting in the classification of its loan accounts as non-performing assets.[2]
Following the defaults, Jammu and Kashmir Bank filed an application under Section 7 of the Insolvency and Bankruptcy Code, 2016 (IBC) before the NCLT, Mumbai, seeking initiation of the corporate insolvency resolution process against EIL.
On 28 August 2024, the NCLT admitted the Section 7 application, appointed an Interim Resolution Professional and declared a moratorium under Section 14 of the IBC.[3] The appellant challenged this decision before the National Company Law Appellate Tribunal (NCLAT).
Before the NCLAT, the appellant argued, among other things, that the liabilities had been affected by a scheme of demerger and subsequent amalgamation. She also contended that a renewed sanction letter dated 18 November 2017 did not mention the guarantee and therefore the guarantee should be regarded as having been relinquished.
The NCLAT rejected the appeal on 11 September 2025. It relied, inter alia, on Clause 8 of the guarantee deed, which provided that the guarantee would not be terminated or prejudiced by absorption or amalgamation of the guarantor company. The NCLAT also referred to several judicial authorities that had apparently been relied upon by the NCLT.[4]
The matter then reached the Supreme Court.
- LEGAL ISSUES
The principal legal issues before the Supreme Court were:
- Whether a judicial or adjudicatory decision based upon fake, non-existent or AI-hallucinated legal precedents can be regarded as a valid decision in law.
- Whether reliance upon AI-generated and unverified judicial authorities compromises the integrity of the judicial process.
- What responsibility rests upon advocates and judges when AI tools are used for legal research and decision-making?
- What directions should be issued to prevent the use of fabricated AI-generated precedents in judicial proceedings?
- Whether the NCLT and NCLAT decisions in the underlying insolvency proceedings could be sustained despite their reliance upon such material.
The case therefore went beyond the immediate insolvency dispute. Its larger concern was the integrity of adjudication in an era in which generative AI is increasingly being used by lawyers, judges, tribunals and other legal professionals.
- ARGUMENTS PRESENTED
(A)Petitioner/Appellant’s Arguments
The appellant, through senior counsel Ms Madhavi Divan, pointed out that the judicial authorities relied upon by the NCLT and subsequently reproduced by the NCLAT were fake or non-existent and were probably AI-generated.[5]
The appellant further demonstrated that even where the citation referred to a genuine judgment, the paragraphs attributed to those judgments could not be located in the relevant law reports. Therefore, the problem was not merely an inaccurate citation but the creation of fictional legal material that was subsequently treated as authentic precedent.
The appellant also advanced arguments concerning the underlying insolvency dispute. In particular, she challenged the continuation of liability under the corporate guarantee following corporate restructuring and relied upon the renewed sanction letter, which allegedly did not contain the guarantee.[6]
The Supreme Court, however, ultimately did not decide the underlying merits because it considered the adjudicatory process itself to have been fundamentally compromised.
(B) Respondents’ Arguments
The respondents opposed the appellant’s submissions on the merits. Jammu and Kashmir Bank had relied upon the corporate guarantee and maintained its claim against the corporate debtor.
Importantly, the Bank subsequently filed an affidavit before the Supreme Court stating that the allegedly fabricated judgments relied upon by the NCLT had not been cited by its counsel. This indicated that the fake authorities had apparently been discovered through the adjudicating authority’s own research rather than being supplied to it by the respondent’s lawyers.[7]
The Supreme Court’s observation on this point is particularly important because it demonstrates the limits of assuming that every authority appearing in a judicial decision has necessarily been verified by counsel or by the court.
- COURT’S REASONING AND ANALYSIS
The Supreme Court began by recognising the transformative potential of AI. It acknowledged that AI has acquired the capacity to assist, and in certain circumstances substantially substitute for, human effort in both routine and intellectual tasks.[8]
The Court nevertheless identified a fundamental distinction between using technology as a tool and allowing technology to substitute for human judgment. Traditional technologies have been incorporated into judicial systems without fundamentally replacing the reasoning process. AI is different because generative systems can produce apparently coherent reasoning, legal authorities and conclusions. The danger, therefore, is not merely technological error but delegation of intellectual responsibility.[9]
The Court stressed that legal adjudication requires the ability to distinguish truth from falsehood and fact from fiction. Judicial decision-making is not merely an exercise in retrieving information. It involves interpretation, evaluation, balancing competing values and exercising judgment.
The most significant part of the judgment concerns AI hallucinations. The Court explained that AI systems may generate non-existent or fabricated material. From the perspective of the legal system, the Court considered the use of such fabricated material as precedent particularly dangerous because judicial decisions depend upon an accurate understanding of authoritative legal sources.
The Court consequently adopted a “zero-tolerance” approach to producing, citing or relying upon AI-generated precedents without verification. It held that an advocate who cites such material without verification commits professional misconduct, while reliance upon such material by a judge constitutes a serious lapse.[10]
Most significantly, the Court held that where fake or hallucinated material enters the decision-making process, the resulting decision is no decision in the eyes of law, irrespective of whether the fabricated material directly or indirectly affected the ultimate conclusion.[11]
The Court also recognised that merely declaring a prohibition would not be sufficient. There must be institutional accountability. Accordingly, it directed the Bar Council of India (BCI) to constitute a committee to consider the problem, formulate guiding principles and prescribe disciplinary consequences for advocates who submit fake or hallucinated material as legal precedent.[12]
The judgment therefore adopts a balanced position: AI may be used, but legal professionals remain responsible for verifying its outputs.
- JUDGMENT AND RATIO DECIDENDI
The Supreme Court set aside the orders of both the NCLT dated 28 August 2024 and the NCLAT dated 11 September 2025. The Section 7 insolvency application was restored to its original number and remitted to the NCLT for fresh consideration on merits.[13]
The Court expressly clarified that it had not expressed any opinion on the merits of the underlying insolvency dispute. The NCLT was required to decide the matter independently according to law. Because the Section 7 application had been pending for a considerable period, the Court directed that it be taken up and disposed of expeditiously, preferably within two weeks. Pending disposal, the parties were directed to maintain the status quo.[14]
- Ratio Decidendi
The central ratio of the judgment can be stated as follows:
A judicial or adjudicatory decision founded upon fake, non-existent or AI-hallucinated legal material is legally unsustainable because reliance upon such material subverts the rule of law and compromises the integrity of adjudication.
The Court also established an important professional principle: AI-generated legal material must be independently verified before being cited or relied upon as precedent.
The judgment therefore creates an important distinction:
- Permissible: Using AI as research, drafting or analytical aid, subject to human supervision and verification.
- Impermissible: Treating AI-generated legal authorities as genuine without checking them against authoritative sources.
The Supreme Court’s official landmark judgment summary similarly identifies the case as concerning the consequences of reliance upon AI-generated fake and hallucinated case law for the sanctity of judicial decision-making. (Supreme Court of India)
- CRITICAL ANALYSIS
(A) Significance of the Decision
The judgment is one of the clearest judicial responses to the emerging risks of generative AI in legal practice. Its importance extends beyond insolvency law.
First, it protects the integrity of precedent. The doctrine of precedent depends upon courts being able to identify genuine judgments and accurately understand their ratio. If fabricated judgments enter the legal system, subsequent courts may unknowingly rely upon fictional authorities, producing a chain reaction of legal misinformation.
Second, the judgment reinforces professional responsibility. Lawyers cannot avoid responsibility by claiming that an AI system generated the material. The ultimate duty to the court remains with the advocate.
Third, the decision is important for judicial accountability. Judges and tribunals must independently verify authorities instead of assuming that technologically generated material is reliable merely because it appears plausible.
Fourth, the judgment provides an important framework for the future regulation of AI in the legal profession. The Court did not advocate an outright ban. Instead, it supported responsible adoption with a human-in-the-loop model. [15]
(B) Implications and Impact
The impact of the judgment is likely to be significant in several areas.
- Impact on lawyers
Lawyers increasingly use AI for case research, drafting and summarising large volumes of material. After this decision, verification becomes not merely good professional practice but an essential ethical obligation. Every AI-generated citation should be checked against an authoritative database, the official court website or an authentic law report.
- Impact on courts and tribunals
Judicial officers must exercise greater caution when relying upon technology-assisted research. The decision makes clear that the ultimate responsibility for the accuracy of legal reasoning cannot be delegated to software.
- Impact on legal education
Law schools should teach students both the capabilities and limitations of generative AI. Legal research training should increasingly include AI verification skills, source authentication and responsible use of technological tools.
- Impact on access to justice
AI can potentially improve access to legal information by reducing research costs and assisting individuals who cannot afford extensive legal services. However, inaccurate AI-generated law can harm vulnerable litigants. Thus, the challenge is to maximise AI’s benefits without compromising accuracy.
The Court’s approach is particularly timely because AI is already being integrated into legal services internationally. The judgment itself refers to regulatory developments in the United Kingdom and notes the importance of supervision, quality control, confidentiality and safeguards against AI hallucinations.[16]
(C) Critical Evaluation
The judgment is commendable for identifying a real and rapidly developing problem. Its strongest feature is that it does not adopt a technologically reactionary position. The Court recognises that AI can assist adjudication but insists that judicial authority must remain human-controlled.
The zero-tolerance principle is also understandable because judicial precedent occupies a special position in the legal system. A fabricated judgment is fundamentally different from an ordinary factual error. It creates an authority that never existed and may therefore distort the development of law.
However, the judgment also raises questions concerning the practical implementation of its strict rule. The Court states that even an “iota” of fake or hallucinated material can invalidate a decision. Such an approach could potentially be difficult to apply where an erroneous citation appears in a lengthy judgment but has no substantive influence on the final outcome.[17] A more nuanced future framework might distinguish between a material reliance on fabricated authority and an accidental typographical or citation error.
Another challenge is determining responsibility where an AI tool is used by multiple actors—lawyers, research assistants, clerks and judicial staff. Clear institutional protocols will therefore be necessary.
Nevertheless, these concerns do not undermine the central principle of the judgment: technology may assist legal reasoning, but it cannot assume legal responsibility.
- CONCLUSION
Pooja Ramesh Singh v Jammu and Kashmir Bank Ltd. & Anr. is a landmark contemporary judgment because it places the emerging relationship between Artificial Intelligence and the administration of justice within a constitutional and professional framework.
The case originated in an insolvency dispute, but its lasting significance lies elsewhere. The Supreme Court confronted the possibility that fabricated AI-generated authorities could enter judicial decisions and thereby undermine the rule of law. It responded firmly by setting aside the NCLT and NCLAT decisions, declaring that reliance upon fake or hallucinated precedents is incompatible with valid adjudication, and directing the Bar Council of India to develop appropriate safeguards.[18]
The judgment does not reject technological progress. Rather, it advocates responsible technological integration. AI can make legal research faster and more efficient, but it cannot replace the lawyer’s duty to verify, the judge’s duty to reason, or the court’s responsibility to administer justice.
The enduring lesson of the case is therefore simple: AI may assist the law, but it must never become the source of legal truth without human verification. As generative AI becomes increasingly embedded in legal practice, the principles established in this judgment are likely to influence professional ethics, judicial administration, legal education and future AI regulation in India.
- BIBLIOGRAPHY/ REFERENCES:
Primary References
- Pooja Ramesh Singh v Jammu and Kashmir Bank Ltd. & Anr., Civil Appeal No. 11950 of 2025, 2026 INSC 668, Supreme Court of India, judgment dated 2 July 2026. (IBBI)
- Insolvency and Bankruptcy Code, 2016, 7, 14.
- Supreme Court of India, Landmark Judgment Summaries, Pooja Ramesh Singh v Jammu and Kashmir Bank Ltd. (Supreme Court of India)
Secondary References
- Law Times Journal, “Case Summary: Pooja Ramesh Singh v Jammu and Kashmir Bank Ltd.” (20 August 2026). (Law Times Journal)
- The Legal Quorum, “Pooja Ramesh Singh v Jammu and Kashmir Bank Ltd. & Anr.” (26 August 2026). (The Legal Quorum)
- Supreme Court of India, “Supreme Court of India launches a new web page – ‘Landmark Judgment Summaries’” (27 September 2024). (Supreme Court of India)
- Rest references are mentioned and attached as hyperlink in the Legal Research Footnotes and Main content as and when required.
[1] Pooja Ramesh Singh v. Jammu and Kashmir Bank Ltd. & Anr., 2026 INSC 668, paras 1–9 (Supreme Court of India, 2 July 2026). (Indian Kanoon)
[2] Ibid., paras 10–11. (Indian Kanoon)
[3] Insolvency and Bankruptcy Code 2016, ss 7 and 14; Pooja Ramesh Singh, 2026 INSC 668, para 12. (Indian Kanoon)
[4] Pooja Ramesh Singh, 2026 INSC 668, paras 12–13. (Indian Kanoon)
[5] Ibid., para 14. (Indian Kanoon)
[6] Ibid., paras 12 and 14. (Indian Kanoon)
[7] Ibid., para 16. (Indian Kanoon)
[8] Ibid., paras 1–2. (Indian Kanoon)
[9] Ibid., paras 3–5. (Indian Kanoon)
[10] Ibid., para 7. (Indian Kanoon)
[11] Ibid., paras 7 and 17. (Indian Kanoon)
[12] Ibid., paras 8–9. (Indian Kanoon)
[13] Ibid., para 18. (Indian Kanoon)
[14] Ibid., paras 19–21. (Indian Kanoon)
[15] Ibid., paras 1, 5 and 7. (Indian Kanoon)
[16] Ibid., paras 2 and 5; see also the Supreme Court of India’s official landmark judgment summary for the case. (Supreme Court of India)
[17] Pooja Ramesh Singh, 2026 INSC 668, para 7. (Indian Kanoon)
[18] Ibid., paras 17–20. (Indian Kanoon)

