Supreme Court quashes Essel insolvency order over tribunal's use of fake, AI-generated precedents

The Supreme Court of India on Thursday has set aside orders passed by two insolvency tribunals after finding that the tribunal relied on fake and non-existent judicial precedents generated using artificial intelligence (AI) in a landmark judgment involving Essel Infraprojects Ltd.

The apex court quashed the orders of both the National Company Law Tribunal (NCLT) and the National Company Law Appellate Tribunal (NCLAT) that had admitted the firm into corporate insolvency, remanding the matter back to the NCLT for a fresh, fact-based adjudication.

The controversy caught the court’s eye when senior advocate Madhavi Divan, representing Essel Infraprojects Ltd, pointed out that several judgements that the NCLT had relied on either did not exist or contained passages that could not be found in the original judgments.

A bench comprising Justice Pamidighantam Sri Narasimha and Justice Alok Aradhe said courts must adopt a “zero tolerance” approach toward the use of fake, AI-generated documents. The bench added that using such hallucinated material without proper verification severely threatens the integrity of the judicial process.

An independent verification conducted by the Supreme Court confirmed the issue. It found that some citations referred to entirely non-existent judgments, while others cited genuine Supreme Court decisions but included fabricated paragraphs attributed to the court.

The case stems from insolvency proceedings initiated by Jammu & Kashmir Bank against Essel Infraprojects Ltd, which had acted as a corporate guarantor for loans given to another firm. Pooja Ramesh Singh, a suspended director of Essel Infraprojects, had challenged the orders admitting the company into the corporate insolvency resolution process under a Section 7 application. The Mumbai bench of the NCLT originally admitted the insolvency plea on 28 August 2024, and the NCLAT subsequently upheld that order on 11 September 2025.

Role of AI in courts

The bench said the case raises a broader concern regarding the role of AI in legal adjudication. While clarifying they were not against using AI in legal practice or the courts, the judges drew a sharp distinction between leveraging AI as an assistive tool and allowing it to replace human judgement, “as it is not just an aid to assist us in our work, but is an alternative to our own thinking, reasoning and even decision-making”.

The court emphasised that while AI can be used as a tool in courts, there must be absolute human control over it at every stage. It also directed the Bar Council of India to form a committee to examine the issue in detail.

Addressing whether citing fake precedents amounts to professional misconduct, Chirag Gupta, associate partner at Alpha Partners, said, “Citing fictitious judgments without verification is misconduct on the part of an advocate, and a judge's reliance on them is a serious lapse. Though the Bar Council of India has been directed to constitute a committee, binding regulations still await formal notification.”

Amitraj Kaushal, an advocate at the Supreme Court of India, noted that fabricated citations and AI-generated errors are far more widespread than currently reported, though many legal professionals are reluctant to admit it publicly.

“I see it across the board, from junior associates rushing to meet deadlines to seasoned practitioners using it for first drafts of arguments. The problem is not that lawyers are using AI. The problem is that they are using it without verifying the output. AI tools hallucinate and can generate case names, citations and even legal propositions that sound completely plausible but simply do not exist,” he said.

This is not the first time Indian courts have flagged AI-generated inaccuracies. In January 2026, the Andhra Pradesh High Court cautioned trial court judges after discovering a lower court had cited four non-existent case laws produced by an AI tool. Similarly, in 2025, the Supreme Court found that a filed rejoinder contained entirely fabricated citations, while the Bombay High Court encountered a tax assessment order that referenced fictitious judicial precedents.

In response to this growing trend, the Supreme Court released draft regulations in June 2026. While the rules permit AI tools for administrative and research functions such as legal research, drafting, translation, transcription, and case management, they strictly prohibit their use in judicial decision-making or determining case outcomes. The draft explicitly warned against the risks of AI hallucinations, inaccurate citations, and the unreliability of automated content.