The Supreme Court of India has delivered a stark warning: blind faith in legal software is a recipe for disaster. On July 2, 2026, a judicial bench comprising Justices Pamidigantam Sri Narasimha and Alok Aradhe vacated a bankruptcy order against Essel Infraprojects Ltd. The reason for the scandal? Lower tribunals had based their verdicts on fictitious precedents helpfully hallucinated by artificial intelligence. The decisions of the National Company Law Tribunal (NCLT) and its appellate body (NCLAT) were not merely overturned but remanded for a complete rehearing. For the business community, the signal is clear: if your legal counsel fails to vet chatbot outputs, your "courtroom victories" become legal liabilities that pose systemic risks to shareholders.
Anatomy of a Judicial Failure
The collapse of legal logic began with a lawsuit from Jammu & Kashmir Bank, which sought to recover debts from Essel Infraprojects as a corporate guarantor. Although the bank successfully navigated two court levels between 2024 and 2025, Essel’s defense counsel, Madhavi Divan, uncovered the deception at the Supreme Court level. It emerged that the tribunal had relied on rulings that never existed, or on genuine cases into which AI had inserted paragraphs of fabricated legal reasoning. The court confirmed the presence of "hallucinations": some citations referred to a void, while others offered distorted interpretations that fundamentally altered the essence of the law.
"This is not merely an auxiliary tool; it is a surrogate for our own thinking, logic, and decision-making process," the judges emphasized, highlighting the risks of unchecked algorithmic implementation.
The court drew a definitive line between AI as a search tool and AI as a replacement for human judgment. The fact that a bankruptcy ruling was annulled due to "garbage data" in the case files puts legal departments worldwide on notice. The Bar Council of India is already forming a committee to study the issue, recognizing that the crisis of confidence in legal documentation could reach critical proportions.
The New Standard of Hard Trust
The Essel Infra case is no isolated incident; it is a systemic trend. Earlier in 2026, the Andhra Pradesh High Court caught judges using four fake statutes generated by a neural network. A similar incident occurred in the Bombay High Court in 2025. In response, the Supreme Court of India issued draft regulations in June 2026 establishing a "zero tolerance" policy: AI is permissible for translation or administrative tasks but strictly prohibited for drafting judicial decisions and conclusions. For business, the "black box" of automated drafting has transformed into a direct compliance threat.
Adopting specialized verification tools is no longer about a preference for technology—it is about survival. The era where a citation in an official document was presumed authentic by default is officially over. Compliance protocols must now integrate a "human-in-the-loop" principle for every line of legal argument. Automation without rigorous oversight—or "Hard Trust"—is a luxury that companies can no longer afford, as a single hallucinated reference can effectively wipe out corporate assets.