Why This Matters Now
In 2026 the Supreme Court set aside orders of the NCLT and NCLAT in the Essel Infraprojects insolvency matter after finding that several cited judgments did not exist and others were real cases stuffed with invented paragraphs. The Court called the intrusion of AI-fabricated material into adjudication catastrophic, ordered the Bar Council of India to frame an AI-use policy, and held that citing fake precedents can be professional misconduct. For an aspirant, this is a sharp GS2 and GS3 case on technology governance, ethics in the justice system and accountability.
The Crux in 60 Words
Generative AI predicts plausible text, not truth, so it invents citations. When such hallucinated precedents entered tribunal orders, the Supreme Court set them aside, called the intrusion catastrophic, adopted zero tolerance for fake authority, and directed the Bar Council to frame norms. The lesson: AI can assist research, but human verification of every citation is non-negotiable, because a fabricated precedent corrupts the reasoning itself.
The Issue, Decoded
| Concept | What it means | Why it matters |
|---|---|---|
| Hallucination | AI generating confident but false output | Manufactures non-existent case law |
| Generative AI | Models that predict likely text | Optimised for fluency, not accuracy |
| Professional misconduct | Citing fabricated authority | Restores the accountability chain |
| Human in the loop | Mandatory human verification | The safeguard AI cannot replace |
The Analysis: Why Fabricated Authority Is So Dangerous
- The record was corrupted, not just delayed. In the Essel Infraprojects matter, tribunal orders rested on judgments that did not exist and real ones distorted with invented text, so the reasoning itself was built on falsehood.
- The flaw is architectural. Large language models are trained to produce fluent, probable text; truth is not their objective, so they hallucinate citations with total confidence.
- Law is uniquely exposed. Legal reasoning stands or falls on the authenticity of authority; a fabricated precedent is not a typo but a defect at the heart of adjudication.
- Accountability must be re-anchored. By treating fake precedents as advocate misconduct and demanding verification, the Court re-establishes who is answerable for what enters a judgment.
Data and Institutions Vault
Carry these into the exam hall.
Case: Supreme Court set aside NCLT and NCLAT orders in the Essel Infraprojects insolvency matter (2026) built on AI-hallucinated precedents; described the intrusion as catastrophic; adopted zero tolerance for fake precedents. Directive: Bar Council of India to frame a policy on responsible AI use; citing fabricated authority can constitute professional misconduct. Bodies: National Company Law Tribunal (NCLT); National Company Law Appellate Tribunal (NCLAT); Supreme Court; Bar Council of India. Concepts: hallucination, generative AI, large language models, human-in-the-loop, algorithmic accountability, explainability. Wider frame: India’s DPDP Act, 2023; principles-based AI governance debates; the idea of AI as assistant, not decision-maker, in public institutions.
The Debate
Argument for strict limits: Generative AI hallucinates authority and cannot verify truth, so allowing it to shape adjudication without human checks corrupts the reasoning and destroys accountability; fabricated citations must be treated as misconduct.
Argument for pragmatic adoption: AI dramatically speeds up research, summarisation and drafting; a blanket ban would deny courts a genuine productivity gain, and the real fault lies with users who fail to verify, not the tool.
Balanced verdict: The pro-AI case concedes the safeguard it disputes. AI is a legitimate research aid but never an authority. The workable position is approved, auditable tools with mandatory human verification and disclosure, so efficiency never comes at the cost of authenticity.
How to Think About This (Transferable Skill)
Distinguish a tool that assists judgment from one that replaces it. For any AI deployment in public life, ask where the human must remain the final authority. Assistance (search, summary, first drafts) is low-risk and reversible; substitution of judgment (deciding rights, sentencing, adjudicating) is high-risk because errors are consequential and hard to reverse. This “assist versus replace” test applies equally to AI in courts, medicine, welfare targeting and policing.
Diagram-in-Words
Generative AI predicts plausible text (not truth) -> hallucinated citations -> enter tribunal orders (Essel Infraprojects) -> reasoning built on non-existent precedents -> Supreme Court sets orders aside, calls intrusion catastrophic -> zero tolerance + Bar Council policy + misconduct liability -> norm: AI assists research, humans verify every citation
The Way Forward
- Frame a clear policy. Let the Bar Council and judiciary jointly define permissible AI use, approved tools and red lines.
- Mandate verification. Require every AI-assisted citation to be checked against primary sources before it enters any filing or order.
- Insist on auditable tools. Prefer approved, legally trained, traceable systems over open consumer chatbots for legal work.
- Require disclosure. Make lawyers declare AI assistance so courts can calibrate scrutiny.
- Penalise fabrication and build literacy. Treat fake authority as misconduct, and train judges and lawyers to understand hallucination and its limits.
The Takeaway Box
Mains angle: Concede AI’s research value, then argue that hallucination makes it unfit to author adjudication; the safeguard is approved tools plus mandatory human verification, disclosure and accountability, as the Supreme Court signalled.
Lift line: “AI can assist justice but must never author it.”
Prelims hooks: Supreme Court set aside NCLT/NCLAT orders in the Essel Infraprojects insolvency matter (2026); AI hallucination; Bar Council of India AI-use policy directive; fabricated precedents as professional misconduct; DPDP Act, 2023.
Ethics / Interview angle: When a tool produces a convincing falsehood, who is morally responsible for relying on it? Where in public decision-making must a human always remain the final authority?
PYQ linkage: UPSC has asked on the impact of emerging technologies and on ethical concerns in the use of AI. This editorial applies both to the justice system.
Connects to: artificial intelligence governance, ethics in public administration, judicial accountability, data protection, technology and society, explainability.
Sources: The Hindu, Supreme Court of India, PIB
Source: When AI Hallucinates Case Law — Ujiyari.com | Free UPSC & State PCS Editorial Analysis
