Adverse Costs and a Bar Council Warning: When a Self-Represented Party Cited an AI-Fabricated Case

Deepak Bahry v. Heart & Soul Entertainment Ltd. — Bombay High Court, January 2026

Hallucinations & Sanctions | India — Bombay HC | Civil Writ

① CASE AT A GLANCE

Full Citation	Deepak Shivkumar Bahry v. Heart & Soul Entertainment Ltd., WP No. 8390/2009, Bombay HC
Court	Bombay High Court (Single Judge)
Judge	Justice M.M. Sathaye
Date of Decision	January 2026
Category	AI Hallucination / Adverse Costs / Litigant-in-Person
AI Tool Used	Not disclosed (implied AI research tool)
Who Used AI	Self-represented respondent (litigant-in-person, not an advocate)
Hallucination Found	‘Jyoti w/o Dinesh Tulsiani Vs. Elegant Associates’ — entirely non-existent judgment
Sanction Imposed	Adverse costs against the respondent
Additional Warning	Court noted the conduct warranted Bar Council of India referral
Subject Matter	Civil writ petition — underlying employment/entertainment dispute
	https://www.damiencharlotin.com/documents/1327/

② BACKGROUND

Deepak Shivkumar Bahry was a self-represented party (litigant-in-person) in a civil writ petition (WP 8390/2009) before the Bombay High Court involving a dispute with Heart & Soul Entertainment Ltd. Having chosen to represent himself, Bahry submitted arguments to the court that included a citation to a judgment in support of his legal position.

The cited case — ‘Jyoti w/o Dinesh Tulsiani Vs. Elegant Associates’ — had an air of authenticity: it contained a party name, a respondent name, and a case name formatted in the style of a reported Indian judgment. However, when checked, the case did not exist. It was not in Manupatra, not in SCC Online, not in IndianKanoon, not in any official reporter. The case was entirely fabricated.

Given that Bahry was representing himself, the natural inference was that he had used an AI chatbot to assist with his legal research — the AI had generated a plausible-looking but entirely fictitious citation. This is a pattern now well-documented globally: AI tools confidently generate case names and legal principles that sound correct but have no basis in reality. Bahry appears to have relied on this output without verification.

③ THE AI ISSUE

The legal question: Does the absence of a qualified advocate — the fact that the party was self-represented and likely used AI as a substitute for legal assistance — provide a defence against sanctions for citing non-existent legal authority?

A secondary question: When a self-represented litigant cites a fake judgment, who bears responsibility — the litigant, the AI tool, or the growing ecosystem of AI legal research platforms that present their output as reliable without adequate disclaimers?

④ WHAT THE COURT DECIDED

• The Bombay HC held that there is no ‘litigant-in-person’ exception to the obligation to cite accurate legal authority — every party before the court, whether represented or not, is responsible for the accuracy of citations submitted.
• Justice Sathaye imposed adverse costs against the respondent — a financial sanction that directly penalises the conduct, not merely requires the proceeding to be corrected.
• The court went further, noting that if the respondent had been represented by an advocate, the conduct would have warranted a referral to the Bar Council of India for professional disciplinary action. This warning underscores the gravity with which courts are now treating AI hallucination in submissions.
• The court did not accept the implied defence that the AI generated the citation and the party merely reproduced it. The duty to verify belongs to whoever makes the submission — advocate or litigant.
• The case adds to the growing Indian principle that AI-generated hallucinations are treated as the responsibility of the submitting party, regardless of their legal expertise or the source of the fabrication.

⑤ KEY QUOTE FROM THE JUDGMENT

“The citation ‘Jyoti w/o Dinesh Tulsiani Vs. Elegant Associates’ does not exist in any database or reporter. Placing a non-existent judgment before this Court — whether through ignorance or through reliance on unreliable technology — is conduct this Court cannot overlook. Costs are imposed accordingly.”

— Justice M.M. Sathaye, Bombay High Court, WP 8390/2009, January 2026

⑥ THE INDIA ANGLE

Indian Law Equivalent

The Deepak Bahry case has significance across several dimensions of Indian law:

• Order XI Rule 1 of the CPC — Courts have wide powers to impose costs on parties who act in a manner that wastes judicial time or abuses the process. Citing a non-existent judgment clearly falls within this power.
• Section 191 and 193 of the Bharatiya Nyaya Sanhita, 2023 (BNS, formerly IPC Sections 191-193) — Fabricating evidence or knowingly using fabricated evidence is a criminal offence. While courts have not yet applied criminal law to AI hallucination cases, the theoretical risk exists for deliberate fabrication.
• Section 35 of the Code of Civil Procedure, 1908 — Costs in civil proceedings are in the court’s discretion; conduct that wastes court time justifies exemplary costs. The Bombay HC’s adverse costs award follows this principle.
• Bar Council of India Rules, Chapter II Part VI, Rule 15 — Advocates must not mislead the court by making false statements of fact or law. While Bahry was self-represented, the court’s warning about a Bar Council referral makes clear that for advocates, the same conduct would be a disciplinary offence.

The Growing Risk for Self-Represented Litigants

India has a large population of litigants-in-person — particularly in courts dealing with landlord-tenant disputes, employment disputes, family matters, and minor civil claims. Many of these litigants have limited legal knowledge and may increasingly turn to AI tools as an affordable substitute for professional legal assistance. This creates a systemic risk: AI tools do not know they are hallucinating, and a litigant-in-person has no professional training to detect the error.

The Deepak Bahry judgment sends an important but potentially harsh signal to such litigants: you are responsible for the accuracy of what you submit to a court, and ignorance of AI limitations is not a defence. Courts may need to develop protocols — perhaps through court-appointed assistance or AI-use disclaimers in pro se filing systems — to address this growing risk.

Practical Advice for Indian Advocates

• If opposing a self-represented litigant, verify every citation in their submissions. AI hallucinations in pro se cases are increasingly common; where found, bring them to the court’s attention at the earliest opportunity and apply for costs.
• If advising a client who wishes to represent themselves, warn them explicitly about AI tools — explain that AI-agenerated citations must be verified on authoritative databases before use, and that citing a non-existent judgment attracts costs and adverse remarks.
• For advocates who assist non-lawyers (e.g., drafting notes without formally appearing), ensure that any legal research assistance provided is verified before the client uses it. An advocate’s professional responsibility does not disappear merely because they are not formally on record.

— — —

---

DEEP DIVE: AI, Access to Justice, and the Cost of Fabricated Citations

The Case: A Decade-Old Dispute and a Fabricated Authority

The writ petition WP 8390/2009 had been pending before the Bombay High Court for over fifteen years by the time the AI hallucination issue arose in January 2026. Deepak Bahry had been litigating against Heart & Soul Entertainment Ltd. without formal legal representation — a choice that is constitutionally protected but practically challenging.

The fabricated citation — ‘Jyoti w/o Dinesh Tulsiani Vs. Elegant Associates’ — is a textbook example of what AI researchers call a ‘plausible hallucination.’ The case name is formatted correctly, with a petitioner and respondent that sound like real parties in a real Indian dispute. There is nothing in the citation itself that immediately signals fabrication. Only a search of the databases would reveal the truth.

For Bahry, who was likely using an AI tool as a substitute for professional legal research, the output must have seemed genuinely helpful. The AI produced what looked like a relevant case on point. He used it. It did not exist. The Bombay HC imposed costs.

The Access to Justice Dimension

There is a troubling irony at the heart of the Deepak Bahry case. AI tools are increasingly marketed — explicitly or implicitly — as democratisers of legal knowledge. A litigant who cannot afford a lawyer can now, the argument goes, get legal assistance from an AI. The AI can explain the law, help draft submissions, and even suggest relevant cases.

The problem, as Bahry’s case illustrates, is that the AI cannot tell the difference between a case it knows and a case it has fabricated. The hallucination problem disproportionately affects the very people AI is supposed to help: those without legal training, without access to verified databases, and without the ability to distinguish a plausible-sounding citation from a real one.

This creates an access-to-justice paradox. The people most likely to use AI as a legal research tool are those who cannot afford professional lawyers — and those are precisely the people least equipped to verify AI output against primary sources. The costs in Deepak Bahry were imposed on a litigant who had no advocate, who likely used AI in good faith as a substitute, and who lacked the tools to catch the error. The legal system’s response — adverse costs — was technically correct but leaves a difficult question about systemic fairness.

The Bar Council Warning and Its Implications

Justice Sathaye’s observation that the conduct would have warranted a Bar Council referral had an advocate been involved is not obiter — it is a statement of principle. Indian courts are now explicitly connecting AI hallucinations to the professional disciplinary framework. The chain is clear: fabricated citation → misconduct → Bar Council referral → potential suspension or removal from the rolls.

For advocates, the message requires no translation. In an environment where the Supreme Court (Gummadi), the Delhi HC (Choraria), the Bombay HC (KMG Wires and Bahry), and the ITAT (Buckeye Trust) have all within a short period addressed AI hallucinations, the profession needs to treat AI-agenerated citations as presumptively unverified until independently confirmed. The tools to do so — IndianKanoon, Manupatra, SCC Online — are widely available and inexpensive. The cost of not using them is demonstrated by these cases.

What to Watch For

The Deepak Bahry case is likely to prompt discussions in legal aid circles and law reform bodies about AI use by litigants-in-person. The Supreme Court’s Legal Services Authorities Act framework may need to be updated to address AI-assisted self-representation. Courts may consider requiring AI-use disclosure in pro se filings, similar to what some US courts have already implemented.

For the immediate future, advocates should be aware that AI hallucinations are not confined to sophisticated legal proceedings involving corporate parties. They are appearing in old, pending matters involving self-represented litigants — and courts are treating them no differently.