Currently Empty: ₹0.00
Myths vs Reality
“The best lawyers of tomorrow will not be replaced by AI — they will be the ones who know exactly where AI ends and legal judgment begins.”
Introduction
Picture two law students on the same evening. The first asks an AI chatbot to draft a legal notice, gets a polished draft in ten seconds, and feels like the future has arrived. The second asks the same chatbot to find “case laws on anticipatory bail,” copies the citations into a court filing without checking them — and later finds out in front of a very unimpressed judge that none of those cases actually exist.
Both scenarios are real, and both are happening in Indian courtrooms right now. As generative AI tools become part of everyday legal work, a genuine question is doing the rounds in law schools, law firms, and even courtrooms: can AI actually replace lawyers? This article separates the popular myths from the legal reality, using the latest developments in Indian and global law — including a landmark ruling by the Supreme Court of India as recently as July 2026.
DID YOU KNOW? As of mid-2026, a global tracker has recorded over 1,300 court proceedings across 106 countries where lawyers submitted fake, AI-generated case citations — with nearly 500 attorneys sanctioned as a result. The very first such case, Mata v Avianca Inc (2023), started when a New York lawyer used ChatGPT for research and it confidently invented six fictional court judgments. |
What Do We Actually Mean by “AI” in Law?
Before busting any myths, it helps to separate two very different things that both get called “AI in law.”
General-purpose AI chatbots (like ChatGPT, Gemini, or Copilot) — trained on huge amounts of internet text, useful for drafting and brainstorming, but not built specifically for legal accuracy.
Purpose-built legal AI tools used by the Indian judiciary itself, such as SUPACE (Supreme Court Portal for Assistance in Court Efficiency) for analysing case records, SUVAS for translating judgments into Indian languages, TERES for courtroom transcription, and LegRAA, the Supreme Court’s own generative research assistant trained on Indian case law.
Both categories are reshaping legal work — but neither has been given any legal recognition to practise law, advise clients, or decide cases. That single fact is the foundation for almost everything that follows.
Myth vs Reality: Busting the Big Claims
MYTH : AI can practise law and give legal advice like a lawyer. REALITY : Under the Advocates Act, 1961, only human advocates enrolled with a Bar Council are recognised to practise law. AI has no legal personality — it cannot be sued, held liable, or disbarred, so it cannot bear professional responsibility for advice given. |
MYTH : AI never makes mistakes because it is a machine. REALITY : AI tools frequently “hallucinate” — they invent fake cases, wrong dates, or fictional quotations with complete confidence. Indian tribunals, trial courts, and even High Court benches have been misled by such fabricated citations. |
MYTH : AI will make lawyers jobless almost overnight. REALITY : Experts tracking Indian legal-tech adoption note that AI is shifting the nature of junior legal work — from routine drafting toward verification, supervision, and client-facing skills — rather than shrinking headcount outright. |
MYTH : Courts in India happily accept AI-generated citations. REALITY : The opposite is true. Courts have quashed orders, imposed costs, and even called reliance on fake AI citations “misconduct.” The Supreme Court held in July 2026 that a decision based on hallucinated judgments is “no decision in the eyes of law.” |
MYTH : Only careless junior lawyers get caught using unverified AI. REALITY : Fabricated citations have slipped past tribunals (NCLT/NCLAT), a Bengaluru income-tax tribunal bench, and multiple High Courts — showing that even experienced benches and counsel can be misled if verification is skipped. |
MYTH : India already has a clear, uniform law regulating AI use by lawyers. REALITY : Regulation is still fragmented: the Kerala High Court issued its own AI policy in 2025, Punjab & Haryana High Court barred judicial officers from using AI tools in 2026, and the Supreme Court’s own AI rules are still in draft/consultation stage. |
Flow Chart: How Should a Responsible Lawyer (or Law Student) Actually Use AI?
STEP 1: Frame the legal research question clearly.
▼
STEP 2: Use an AI tool to generate a first draft or a list of possible case laws.
▼
STEP 3: STOP — do not trust any citation yet.
▼
STEP 4: Independently verify every case name, citation, and quotation on a trusted database (SCC Online, Manupatra, official court website).
▼
STEP 5: Cross-check that the case is still “good law” and actually says what AI claims it says.
▼
STEP 6: Only after verification, use the material in a court filing, article, or opinion — under the lawyer’s own name and accountability.
Difference Between AI Tools and Human Lawyers
Parameter | AI Tools | Human Lawyers |
Legal recognition | None under the Advocates Act, 1961 | Enrolled and regulated by the Bar Council |
Accountability for errors | Cannot be held liable or disciplined | Personally accountable; can face sanctions or disciplinary action |
Speed of research | Extremely fast — seconds to minutes | Slower, but grounded in verified understanding |
Accuracy without checking | Can confidently produce fabricated “facts” | Trained to verify sources and apply judgment |
Courtroom advocacy & persuasion | Cannot appear, argue, or persuade a judge | Core professional skill |
Negotiation & client empathy | No genuine emotional or ethical judgment | Central to client counselling and settlement |
Ethical duties (confidentiality, candour) | No independent ethical obligations | Bound by the Advocates Act and BCI Rules |
Landmark Cases & Real Incidents Every Student Should Know
Case Watch
Mata v Avianca Inc, 678 F Supp 3d 443 (SDNY 2023) — The case that started it all. A US lawyer submitted a brief citing six fictitious cases generated by ChatGPT. When the judge asked for copies, the lawyer even asked the same chatbot to “confirm” the cases were real — and it did. This remains the most-cited global reference point for AI hallucination sanctions.
Gummadi Usha Rani v Sure Mallikarjuna Rao — A Vijayawada trial court relied on four non-existent Supreme Court judgments. On appeal, the Andhra Pradesh High Court initially held that fake citations do not automatically void an order if the legal principle was correct — but the Supreme Court later overruled this lenient approach, holding in July 2026 that a decision built on fabricated judgments is “no decision in the eyes of law” and amounts to misconduct.
Essel Infraprojects insolvency matter — The Supreme Court set aside orders of the NCLT and NCLAT after discovering that of six precedents relied upon, three judgments did not exist at all and three genuine ones contained invented paragraphs, showing that even specialised tribunals are vulnerable.
KMG Wires Pvt Ltd — The Bombay High Court quashed a tax order worth over ₹27 crore after finding it was built on precedents that simply did not exist.
The “phantom paragraphs” incident — In September 2025, a Delhi High Court petition was withdrawn in embarrassment after opposing counsel discovered it quoted paragraphs 73 and 74 of a judgment cited as Raj Narain v Indira Nehru Gandhi, (1972) 3 SCC 850 — a citation that, on verification, did not support the paragraphs the AI tool had invented and quoted. Note for students: this citation should not be confused with the famous basic structure ruling Indira Nehru Gandhi v Raj Narain, 1975 Supp SCC 1 : AIR 1975 SC 2299 (7 November 1975) — a good example of how easy it is to mix up two differently-styled cases between the same parties, which is exactly the kind of error AI tools (and hurried humans) can make.
Practical, Real-Life Examples of AI Already at Work
SUPACE helps Supreme Court judges sift through large volumes of case records before a hearing.
SUVAS has translated tens of thousands of Supreme Court judgments into Hindi and other Indian languages, improving access to justice for non-English speakers.
Adalat AI now transcribes live hearings in thousands of courtrooms across several Indian states, cutting down delays caused by manual note-taking.
A Bengaluru income-tax tribunal bench had to recall an order in a matter worth over ₹669 crore after discovering that four of the precedents it relied on, generated using AI, simply did not exist.
Law firms increasingly use AI for first-level contract review and due diligence — but final sign-off always rests with a qualified lawyer.
Relevance in Today’s Time
India’s courts are sitting on a backlog of over 5 crore pending cases, which experts estimate would take human judges several centuries to clear at the current pace. This is exactly why the judiciary itself is leaning on AI tools like SUPACE, SUVAS, and LegRAA for efficiency.
At the same time, the same wave of technology has created an entirely new professional risk: fabricated citations. For a law student in 2026, this makes verification skill — not typing speed — the real differentiator.
The Supreme Court’s November 2025 white paper on AI and the judiciary makes human verification mandatory at every stage, and the Bar Council of India has itself acknowledged that only human advocates can be held responsible for AI-generated errors.
In short, AI is not a threat to be feared or a shortcut to be blindly trusted — it is a tool that raises the bar for competence. Lawyers who learn to use it responsibly will out-compete those who either ignore it completely or trust it blindly.
Frequently Asked Questions (FAQs)
Q1. Can I cite an AI-generated case law directly in a court filing?
No. Every citation must be independently verified on a trusted legal database before it is used in any filing, article, or opinion.
Q2. What happens if a lawyer files fake AI-generated citations?
Consequences have ranged from heavy costs and monetary sanctions to referral to bar disciplinary authorities and, in serious cases, disqualification from a matter.
Q3. Are AI research tools banned in Indian courts?
No — the judiciary uses its own AI tools extensively. What is required is mandatory human verification, not a ban on the technology itself.
Q4. As a law student, should I avoid using AI altogether?
Not necessarily. AI can genuinely speed up first drafts and research. The non-negotiable rule is: never submit or rely on anything AI gives you without checking it yourself, line by line.
QUICK REVISION BOX
|
Conclusion
AI cannot replace lawyers — but lawyers who use AI responsibly are quietly starting to outpace those who don’t, while those who use it carelessly are ending up in disciplinary proceedings. The myth of AI as either a magic replacement for legal minds or a harmless shortcut both miss the real story: AI is a powerful but unreliable junior assistant that still needs a qualified, accountable human lawyer checking its work at every step.
For law students entering practice in the next few years, the most valuable skill will not be knowing how to prompt an AI tool — it will be knowing exactly when not to trust it.
TECH & CYBER LAW | STUDENT LEGAL ARTICLE SERIES
by Khyatee Parashar on 05/07/2026





