AI Hallucinations in Indian Courts: A Complete Guide to Every Case (2024–2026)

From Trial Courts to the Supreme Court — How India Is Confronting the AI Citation Crisis

Comparative Analysis | India — All Courts and Tribunals | 2024–2026

⚡ INDIA ROUND-UP

AI Hallucinations in Indian Courts and Tribunals — 2024–2026

In less than 18 months, five Indian courts and tribunals — including the Supreme Court — have confronted AI hallucinations: fake judgments cited by tax officers, a judge, a litigant, and the tribunal itself. India is now the world’s most active jurisdiction for AI law hallucination cases.

Why it matters: India’s emerging AI law jurisprudence is more comprehensive than any other country’s — covering advocates, litigants, administrative officers, judges, and tribunals. Every practitioner needs to know these five cases.

Category: Comparative Analysis / India Hallucination Cases | Jurisdiction: India — All Courts | Read time: 10 min

THE FIVE CASES AT A GLANCE

Case	Court / Tribunal	Date	Who Used AI	Outcome
Buckeye Trust v. PCIT	ITAT Bangalore	Dec 2024	Tribunal (implied)	Order WITHDRAWN & re-heard
KMG Wires v. ITO	Bombay HC	Oct 2025	Assessing Officer	Rs. 27.91 Cr assessment QUASHED
Jeetmal Choraria v. UoI	Delhi HC	Nov 2025	GST Authority	Show Cause Notice SET ASIDE
Deepak Bahry v. Heart & Soul	Bombay HC	Jan 2026	Self-represented litigant	Adverse costs; Bar Council warning
Gummadi Usha Rani v. Sure MR	AP HC + Supreme Court	Jan–Feb 2026	Trial Court Judge	Misconduct; AG/SG/BCI on notice

HOW INDIA GOT HERE: THE 18-MONTH TIMELINE

India’s AI hallucination crisis in the legal system did not appear overnight. It reflects a broader pattern of AI adoption across professional contexts — driven by the accessibility of AI tools, the volume of litigation, and the pressure on legal professionals to work faster and at lower cost.

The Buckeye Trust ITAT order of December 2024 was the first documented incident — though it was only retrospectively identified as an AI hallucination case after the order was reportedly withdrawn. The KMG Wires and Jeetmal Choraria cases followed in quick succession in late 2025, establishing two key principles: that AI hallucinations in quasi-judicial proceedings void the proceeding, and that courts will verify citations themselves. The Deepak Bahry and Gummadi cases in early 2026 completed the picture — with the Supreme Court’s intervention in Gummadi elevating the issue to a matter of national institutional concern.

WHO IS USING AI — AND WHY IT MATTERS

What makes India’s AI law jurisprudence distinctive globally is the breadth of who has been found to use AI-generated content in legal proceedings:

Administrative quasi-judicial officers (Assessing Officers under the Income Tax Act; GST enforcement officers) — KMG Wires and Jeetmal Choraria.
Self-represented litigants — Deepak Bahry.
Judicial officers of subordinate courts — Gummadi AP HC and SC.
Appellate tribunal members (implied) — Buckeye Trust ITAT.

In contrast, in the United States — where AI hallucination cases began earlier, with Mata v. Avianca in 2023 — virtually all documented cases involve advocates submitting AI-generated citations to courts. No US judge or tribunal member has been publicly identified as using AI-generated hallucinations in a formal order.

India’s pattern is therefore both more alarming and more instructive. The AI risk does not sit only with advocates — it sits throughout the entire legal system. And the Indian judiciary’s response, particularly through the Supreme Court’s intervention in Gummadi, is correspondingly more systemic.

THE EMERGING PRINCIPLES: WHAT THE CASES ESTABLISH

1. No Good Faith Defence

Across all five cases, the courts have consistently rejected the argument that AI hallucinations can be excused on the ground of good faith. Whether the user is an AO, a judicial officer, or a self-represented litigant, the duty to verify is absolute. The Gummadi AP HC was most explicit: AI lacks consciousness and moral reasoning, and cannot substitute for professional verification.

2. Fabricated Citations Void the Proceeding

An assessment order (KMG Wires), a GST SCN (Jeetmal Choraria), a court order (Gummadi), and a tribunal order (Buckeye Trust) have all been either quashed, set aside, or withdrawn because they cited non-existent legal authority. The principle is clear: there is no ‘harmless error’ doctrine for fabricated citations in India.

3. Courts Will Verify

The Delhi HC physically retrieved books from its library to verify citations (Jeetmal Choraria). The Bombay HC checked legal databases (KMG Wires). The AP HC accepted petitioner’s evidence of non-existence (Gummadi). Indian courts are not passive recipients of citations — they are actively verifying.

4. Misconduct, Not Error

The Supreme Court’s declaration in Gummadi that AI hallucination by a judicial officer is ‘misconduct’ — not a correctable error — sets a bright line for disciplinary purposes. This principle, once established at the Supreme Court level, is likely to be applied by High Courts in other contexts: an advocate who cites a fabricated case may similarly be held to a ‘misconduct’ standard rather than a ‘bona fide error’ standard.

5. National Guidelines Are Coming

The Supreme Court’s notice to the AG, SG, and Bar Council in Gummadi is a clear signal: binding national guidelines on AI use in legal proceedings are anticipated. Advocates, judicial officers, and administrative authorities should prepare for a regulatory framework that will likely require: (a) disclosure of AI use; (b) mandatory verification of citations; (c) professional consequences for unverified AI submissions.

THE INDIA ADVANTAGE: WHY THESE CASES MATTER FOR YOUR PRACTICE

India’s AI law jurisprudence is already more developed than most practitioners realise. Every advocate in India — from junior associates to senior counsel, from trial court practitioners to ITAT specialists — now needs to operate against the backdrop of these five cases. The practical implications are specific and immediate:

Citation verification is now a professional obligation, not a good practice. Any citation you submit to a court, tribunal, or quasi-judicial authority must be independently verified before submission. This is not optional — it is now a standard of professional conduct with disciplinary consequences.
When receiving orders and SCNs, check every citation. An order or notice that contains even one hallucinated citation may be void — and may be challenged on that ground alone, regardless of the substantive merits.
The Bar Council of India will likely issue guidance following the Supreme Court proceedings in Gummadi. Watch for circulars in 2026 that formalise the AI verification obligation.
GST practitioners face particular exposure: the Jeetmal Choraria case gives taxpayers a tool to challenge SCNs on citation grounds at the outset. This has significant implications for the entire GST enforcement calendar.
Income tax practitioners — both for taxpayers and Revenue — face similar exposure: KMG Wires at the assessment level and Buckeye Trust at the ITAT level demonstrate that AI hallucinations can vitiate proceedings at every stage of the income tax hierarchy.

WHAT INDIA CAN LEARN FROM THE US EXPERIENCE

The United States confronted its AI hallucination problem in 2023 with Mata v. Avianca, followed by a series of escalating sanctions culminating in the $110,000 Oregon penalties in 2025. The US response has been court-specific: individual courts issuing standing orders requiring AI disclosure and certification, rather than a national regulatory framework.

India has an opportunity to do better. The Supreme Court’s proactive intervention in Gummadi, combined with the BCI’s inclusion in the notice, creates the conditions for a comprehensive national framework rather than a patchwork of individual court orders. If India can establish a single national standard — requiring AI disclosure, citation verification, and professional consequences — it will have created the most advanced AI law framework of any major jurisdiction.

The five cases documented in this analysis are not the end of India’s AI law story. They are the beginning. Every court, every tribunal, every GST officer, every AO, and every advocate is now on notice: the era of unverified AI citations in Indian legal proceedings is over.

QUICK REFERENCE: ALL FIVE CASES

Buckeye Trust v. PCIT	ITAT Bangalore, Dec 2024. Tribunal order reportedly withdrawn over suspected AI hallucinations: wrong HC, wrong party name, repealed provision. Blog: /blog/buckeye-trust-itat-bangalore-order-withdrawn-ai-hallucinations/
KMG Wires v. ITO	Bombay HC, Oct 2025. AO cited 3 fake judgments; Rs. 27.91 Crore income tax assessment quashed. Blog: /blog/kmg-wires-income-tax-assessment-quashed-ai-fake-judgments/
Jeetmal Choraria v. UoI	Delhi HC, Nov 2025. GST SCN with 1 fake + 1 wrong citation; HC physically checked library books; SCN set aside. Blog: /blog/jeetmal-choraria-delhi-hc-gst-notice-ai-hallucinations/
Deepak Bahry v. Heart & Soul	Bombay HC, Jan 2026. Pro se litigant cited fake ‘Jyoti Tulsiani’ case; adverse costs; Bar Council warning. Blog: /blog/deepak-bahry-bombay-hc-adverse-costs-ai-fabricated-citation/
Gummadi Usha Rani v. Sure MR	AP HC + SC, Jan–Feb 2026. Trial court judge cited 4 AI-generated fake cases; SC declared it misconduct; AG/SG/BCI on notice. Blog: /blog/gummadi-usha-rani-judge-ai-hallucination-supreme-court/