Supreme Court addresses precedents created by AI
NEW DELHI: The Supreme Court on Thursday said that fake and non-existent precedents created by artificial intelligence pose the same threat to the justice system that methyl isocyanate poses to human life – “invisible, insidious and catastrophic by the time anyone notices”.
The court was referring to a ruling by the National Company Law Tribunal which cited several hallucinatory precedents.
The Supreme Court added that such materials pollute the judicial process and strip the judicial decision of its “lifeblood.”
Cautioning against the use of AI in court proceedings, a bench of Justices PS Narasimha and Alok Aradhe set aside the rulings of the NCLT and the National Company Law Appellate Tribunal (NCLAT) in an insolvency dispute after finding that the NCLT relied on non-existent, spurious and hallucinogenic precedents generated by AI tools and that the NCLAT failed to detect error while upholding the NCLT order.
The proceedings in court strongly suggested that the NCLT may have relied on AI tools to write its judgement, although the bench chose not to delve into this matter.
The bench’s reference to methyl isocyanate goes back to the 1984 Bhopal gas tragedy that claimed thousands of lives in the world’s most serious industrial accident.
The bench also said that all courts must adopt a “zero tolerance” policy for citing or relying on “false, non-existent and hallucinogenic material” generated through artificial intelligence.
It directed the Bar Council of India (BCI) to constitute a committee to study the issue of lawyers presenting fake or hallucinogenic material generated by artificial intelligence before courts as if it were a real precedent. It asked the BCI to establish guidelines and determine disciplinary consequences for violations.
The Supreme Court said: “For us, that is, for those working in the field of adjudicating disputes, this by-product of artificial intelligence, the production of false, non-existent, hallucinogenic materials and their use as precedents in the law, is like the release of methyl isocyanate into the field of law and justice: invisible, insidious, and catastrophic at the time when no one notices.”
The Supreme Court also said that any lawyer who cites hallucinogenic judgments commits professional misconduct, and that judges who rely on such judgments or precedents are doing so gravely wrong.
The court said that any decision tainted by even “a shred” of these provisions “is not a decision in the eyes of the law” and must be overturned, regardless of whether the false precedent influenced the final outcome.
The court was hearing an appeal filed by the suspended director of a private company, challenging the NCLT’s orders admitting the company in the corporate insolvency resolution process on the basis of a Section 7 application filed under the Insolvency and Bankruptcy Code by the Jammu and Kashmir Bank.
NCLT Mumbai, on August 28, 2024, recognized the insolvency declaration and in September 2025, NCLAT upheld the order.
However, senior advocate Madhavi Divan, who appeared for the appellant, told the apex court that many of the citations or judgments relied upon by the NCLT either do not exist or contain paragraphs not recorded by any legal report.
The court then conducted its own examination and found multiple errors. Some citations referred to provisions that did not previously exist. Others cited real cases but attributed fabricated clips to them. One citation even bore the name of a completely different Supreme Court decision.
The bank told the court that its lawyer did not cite these things and that the judiciary may have cited them through its own research.
The court noted that the fictitious precedents had escaped scrutiny even before the NCLAT. She also said that courts cannot realistically verify the validity of every judgment cited by lawyers and that the justice system depends on lawyers placing real powers before the court.
She said that while technology had always helped dispense justice, artificial intelligence was different because it could “replace human thinking, logic and decision-making.”
The court said that the ability to distinguish truth from falsehood, right from wrong, justice from injustice, and “dharma and adharma” develops through discipline and life experience, and that abandoning the exercise of independent thought would erode the basis of human judgment.
She added that courts alone cannot solve the problem and that “absolute and complete control” over AI will require public policy and enforceable rules. She added that the real guarantee lies in the willingness of judges and lawyers to use technology with “restraint and verification.”
