In a landmark judgment that is likely to redefine the relationship between artificial intelligence and the administration of justice in India, the Supreme Court has unequivocally declared that citing AI-generated fake or “hallucinated” judicial precedents without verification constitutes professional misconduct on the part of advocates, while any judicial decision founded upon such fabricated authorities is legally void and incapable of sustaining judicial scrutiny. The ruling marks the first authoritative pronouncement by the apex court laying down a comprehensive institutional framework for the responsible use of artificial intelligence in litigation, while simultaneously drawing a clear constitutional distinction between the legitimate use of AI as an assistive technological tool and the impermissible reliance upon fabricated legal authorities. Emphasising the sanctity of judicial reasoning, the Court called for a “zero-tolerance” approach towards the production, citation or reliance upon AI-generated fake precedents, observing that the integrity of adjudication cannot be compromised by technological shortcuts or unverified digital outputs.
The judgment was delivered by a Bench comprising Justice P. S. Narasimha and Justice Alok Aradhe while deciding an appeal arising from insolvency proceedings initiated under Section 7 of the Insolvency and Bankruptcy Code by Jammu and Kashmir Bank Limited against Essel Infraprojects Limited. The controversy assumed extraordinary significance after it emerged that both the National Company Law Tribunal (NCLT) and the National Company Law Appellate Tribunal (NCLAT) had relied upon several judicial authorities that simply did not exist. During the hearing, Senior Advocate Madhavi Divan, appearing for the suspended director of the corporate debtor, demonstrated that a number of precedents cited in support of the impugned orders—including purported Supreme Court judgments—could not be traced in any recognised legal database. An affidavit placed before the Court confirmed that the cited authorities were entirely fictitious and bore the characteristics of AI-generated “hallucinations”, thereby raising serious concerns regarding the authenticity of the adjudicatory process itself.
Accepting these submissions, the Supreme Court set aside the orders of both the NCLT and the NCLAT, holding that a judicial determination tainted by fabricated precedents is “no decision in the eyes of law.” In one of the strongest observations made by the Court on the subject of artificial intelligence, the Bench ruled that even if only “an iota” of fake or hallucinated material enters the judicial reasoning process, the resulting decision cannot be permitted to stand because it fundamentally violates the sanctity of adjudication. The Court made it clear that the issue was not merely one of erroneous legal reasoning but one striking at the very legitimacy of the judicial process. Unlike an ordinary mistake in appreciation of law, reliance upon non-existent precedents undermines public confidence in the administration of justice and therefore warrants invalidation of the entire decision irrespective of whether the fabricated authority directly influenced the final outcome.
The Court then proceeded to formulate what may become one of the defining ethical principles governing artificial intelligence in Indian legal practice. It categorically held that an advocate who cites AI-generated judgments without first independently verifying their authenticity commits professional misconduct. Simultaneously, it observed that it would amount to a “serious lapse” on the part of judges or adjudicating authorities if they rely upon such fabricated material while determining disputes. The Bench emphasised that courts and tribunals traditionally repose implicit trust in members of the Bar while considering authorities cited during arguments. If advocates themselves fail to discharge their professional obligation of verification, the burden cannot realistically be shifted onto judges to authenticate every precedent cited before them. Such a situation, the Court observed, would seriously impair the efficiency of the judicial system while simultaneously undermining the trust upon which courtroom practice has historically functioned.
One of the notable aspects of the judgment is the Court’s concern regarding the functioning of appellate institutions. The Bench questioned how the fabricated citations escaped scrutiny not only before the NCLT but also before the NCLAT, which functions as the first statutory appellate authority. The Court remarked that if imaginary precedents can travel unnoticed through multiple levels of adjudication, the consequences for the rule of law become deeply institutional rather than individual. This observation broadens the debate from professional ethics to systemic judicial accountability, suggesting that every participant in the adjudicatory process from counsel to tribunals and appellate bodies—shares responsibility for maintaining the authenticity of legal sources relied upon in judicial decision-making.
Recognising that the issue extends beyond the facts of the present case, the Supreme Court directed the Bar Council of India to examine the matter with utmost seriousness and formulate guiding principles governing the use of artificial intelligence by advocates. The Court directed the regulatory body to prescribe norms requiring verification of AI-generated material and to identify appropriate disciplinary consequences for violations of such standards. The direction assumes particular significance because the Advocates Act, 1961 already casts professional and ethical obligations upon advocates to maintain competence, diligence and candour before courts. By linking AI-generated fake citations with professional misconduct, the Supreme Court has effectively integrated emerging technological challenges into the existing framework of legal ethics rather than treating them as entirely new forms of misconduct.
Importantly, the Bench was careful to clarify that its judgment should not be interpreted as hostility towards artificial intelligence itself. On the contrary, the Court expressly recognised the legitimate utility of AI in legal research, document review, drafting assistance, translation, summarisation and other administrative functions. The prohibition, according to the Court, is confined to presenting fabricated or hallucinated AI outputs as authentic judicial precedents. This nuanced distinction reflects a balanced judicial approach that embraces technological innovation while reaffirming that responsibility for legal accuracy always rests with human professionals rather than algorithms. In doing so, the Court avoided the extremes of either uncritical technological optimism or blanket prohibition, instead insisting upon meaningful human supervision over every AI-assisted legal output.
The ruling also fits within a broader international trend concerning judicial regulation of generative AI. Across several jurisdictions, courts have sanctioned lawyers for filing pleadings containing fictitious authorities generated by AI systems, most notably following the widely discussed Mata v. Avianca litigation in the United States, where counsel relied upon entirely fabricated case law produced by an AI chatbot. Similar disciplinary proceedings have since emerged in other jurisdictions, reinforcing a global consensus that while AI may enhance legal productivity, it cannot dilute the lawyer’s professional duty to verify every authority cited before a court. The Indian Supreme Court’s decision therefore aligns domestic jurisprudence with an increasingly accepted international principle that technological convenience cannot override ethical responsibility.
From a jurisprudential perspective, the judgment significantly expands the doctrine of procedural integrity within judicial decision-making. Traditionally, appellate courts interfere with judgments on grounds such as jurisdictional error, perversity, violation of natural justice or misapplication of legal principles. By holding that reliance upon fabricated judicial authorities itself renders a decision legally non-existent, the Supreme Court has introduced a distinct category of institutional invalidity. The emphasis is no longer confined to whether the substantive conclusion is correct; equal importance is now attached to the authenticity of the reasoning process through which that conclusion is reached. In effect, the Court has reaffirmed that the legitimacy of adjudication depends not merely upon correct outcomes but upon demonstrably authentic legal reasoning supported by verifiable authorities.
The decision is also likely to reshape legal education and professional training in India. As generative AI rapidly becomes integrated into law firms, chambers and court practice, young lawyers increasingly rely upon AI-assisted research tools for preparing pleadings and identifying precedents. The judgment sends an unequivocal message that AI may function as an assistant but can never become a substitute for independent legal research or professional judgment. Every authority cited before a court must continue to be verified through recognised legal databases, official law reports or authenticated judicial repositories. Failure to do so is no longer merely negligent drafting, it now carries the potential consequence of professional disciplinary proceedings.
Beyond the immediate insolvency dispute, the ruling establishes a constitutional benchmark for the future governance of artificial intelligence within India’s justice delivery system. By simultaneously protecting technological innovation and preserving the integrity of adjudication, the Supreme Court has articulated an enduring principle that is likely to influence future regulation of AI across courts and tribunals. The judgment makes it abundantly clear that while artificial intelligence may transform legal practice, it cannot be permitted to transform the foundational values upon which judicial legitimacy rests. In an era where algorithms increasingly assist human decision-making, the Court has reaffirmed that accountability, authenticity and verification remain exclusively human responsibilities. The decision therefore stands not merely as a warning against AI hallucinations but as a broader reaffirmation that the rule of law can never be outsourced to machines.
