I. Introduction

In 2023 a lawyer in New York filed a brief that cited six earlier cases. The judge asked for copies. The lawyer could not give them, because the cases were not real. ChatGPT had been used for the research, and it had made up case names, courts and years that looked genuine. The story went around the world. It left courts, law schools and bar councils with a hard question. What is AI doing inside the legal system, and are lawyers ready for it?

India soon had its own case. In July 2026 the Supreme Court set aside orders of the National Company Law Tribunal (NCLT) and the National Company Law Appellate Tribunal (NCLAT). The orders relied on AI-generated precedents that were fake or wrongly quoted. The Court said that citing such material without checking it is misconduct by an advocate. It also asked the Bar Council of India to form a committee to frame guidelines. A month earlier, the Supreme Court’s AI Committee had released draft rules on AI in courts. These are good steps, but only first steps. Many questions are still open. How should client data be handled? Who is to blame when an AI tool gets something wrong?

The article is arranged as follows. Section II explains what AI is and why it fits law so well. Section III shows how lawyers use it today, and Section IV names some tools. Section V covers the problems. Section VI sets out Indian law, and Section VII looks briefly at other countries. Section VIII compares benefits and risks, Section IX looks ahead, and Section X concludes.

II. What AI Means for Lawyers

A. What Is Artificial Intelligence?

In simple words, AI is software that does work which used to need a human mind. It can spot patterns, understand language, make predictions and write new text. Three kinds matter most for lawyers.

• Machine learning means a system learns from examples instead of fixed rules. A contract tool that gets better at finding key clauses after reading thousands of agreements is one example.
• Natural language processing lets a computer work with everyday language. A research tool that finds a judgment because it matches the idea of your question, and not only the words, uses it.
• Generative AI is the type most people know from ChatGPT. It writes new text based on patterns in huge amounts of data. It can draft a contract or summarise an article, but it can also be wrong while sounding sure.

B. Why This Matters for Law

Law is mostly reading and writing. A lawyer spends a lot of time on judgments, drafts and papers, and that is where AI is strongest. A junior advocate may need three days for a constitutional question. An AI tool can give an outline in half an hour. The saved time can go to thinking, to clients and to preparing for court.

There is a catch, though. A young lawyer who accepts AI answers without question never learns to research well. That skill may be missing when it matters most.

III. How AI Is Used in Legal Practice Today

A. Legal Research

Lexis+ AI and Westlaw Precision, and Indian services like SCC Online and Manupatra, now let lawyers ask questions in ordinary language. They no longer need keyword strings. A lawyer can describe a problem, for example when High Courts have granted interim relief in intellectual property disputes against foreign defendants. The system then brings up relevant decisions from different courts and years. India has a very large body of case law, so this saves a lot of time.

B. Contract Drafting and Review

Tools like Kira, Luminance and ContractPodAi read a contract, pick out the key clauses, flag unusual terms, compare them with market practice and summarise the risks. A human team cannot work that fast. For routine papers such as non-disclosure agreements, employment contracts and vendor agreements, generative AI can write a first draft, which the lawyer then edits. Firms that handle mergers and acquisitions have started using these tools for the many contracts that come up in due diligence, especially when time is short.

C. Predicting Case Outcomes

AI can study past judgments, including how particular judges decide and which arguments tend to work, and then estimate how a case may go. Lex Machina in the United States is a well-known example. India is behind here. Digital case data is less complete, especially from lower courts, and a prediction is only as good as its data. Even so, a rough pattern can help. A tribunal bench that often sides with the revenue authorities is one example. Knowing this helps a lawyer plan strategy or decide whether to settle.

D. Reviewing Large Numbers of Documents

Big commercial disputes and regulatory investigations can involve huge numbers of documents. AI review tools sort them by relevance and finish in hours what a human team would take weeks to do. Research suggests that this technology-assisted review can match a full manual review, and sometimes beat it. It is used more and more in international arbitration involving Indian parties.

E. Help for Clients

Chatbots on law firm and legal aid websites can answer simple questions about procedure, guide people through filing a complaint and list the papers they need. Think of a person going to a consumer forum for the first time. Or think of a domestic violence survivor looking for information late at night, when every office is closed. A chatbot can help them. Some legal-technology start-ups are also building chatbots in regional Indian languages.

F. Compliance Monitoring

Banks, pharmaceutical companies and other heavily regulated businesses must follow rules that keep changing and differ from place to place. AI compliance tools read new regulations, amendments and court orders as they come out, flag the important ones and show how they affect the business. India issues many circulars and often amends subordinate legislation, so this kind of tracking is very useful.

IV. Common AI Tools

The table below names some of the main AI tools and how they relate to Indian practice. Lawyers are expected to know these names more and more, even if they never use the tools.

Tool	Main use	Relevance for India
ChatGPT (OpenAI)	Drafting, research summaries, structuring arguments	Widely used for early research. Not reliable for citations
Harvey	AI assistant for law firms, covering documents and strategy	Mostly used by large international firms
Lexis+ AI	Legal research with AI-written summaries	Offered in India through LexisNexis
CoCounsel (Thomson Reuters, formerly Casetext)	Case analysis and legal memos	Mainly relevant to cross-border matters
Kira	Contract review and due diligence	Used in mergers and acquisitions work
Luminance	Document review for disputes and contracts	Growing use in commercial work
SCC Online / Manupatra	Indian case law databases with AI search	Most useful for Indian legal research
CaseMine / SpotDraft	Indian legal research and contract tools	Built around Indian law and practice

Still, no tool can replace the judgment of a qualified lawyer. Whatever an AI system produces, whether research, a draft or a suggested strategy, must be checked before anyone relies on it. The tool is only an instrument. The lawyer remains responsible for how it is used.

V. Legal and Ethical Challenges

A. Fake Information (Hallucination)

Large language models are built to sound right, not to be right. If you ask one about a case that does not exist, it may invent a case name, a citation and a court, and present them as confidently as a real authority. This is what happened in Mata v Avianca. Fake citations went into court filings and the lawyers were sanctioned. India has now seen the same thing, and the Supreme Court’s 2026 ruling treats unchecked AI citations as professional misconduct (n 2).

The rule for lawyers is short. Nothing produced by AI should go into a court filing or a formal opinion until someone has checked it against the original judgment or statute. Asking the same AI tool if its answer is right does not count as a check.

B. Privacy and Confidentiality

Say a lawyer uploads client documents to an AI platform. Where does the data go? Who can see it? Will it be used to train the next version of the model? Indian advocates must keep client communications confidential. This duty comes from the Advocates Act 1961, the Bar Council of India Rules and the law on privileged communications. It is not yet settled whether uploading client material to outside AI services, especially services hosted abroad, fits with this duty. Until it is clear, the safer way is to take the client’s informed consent first and to keep sensitive material out of public AI tools.

C. Bias

AI learns from past data, so it can pick up the biases in that data and even increase them. Imagine a prediction tool trained mostly on English-language decisions of the High Courts and the Supreme Court. Its results may be unreliable for lower court disputes, or for communities that hardly appear in the data. This is not only a technical problem. It affects access to justice, and the profession has not paid enough attention to it. In my view it is one of the least discussed problems in the whole debate.

D. Who Is Responsible?

When a lawyer makes a mistake, the client can complain to the disciplinary body or sue for negligence. When an AI system makes the mistake and the lawyer relies on it without checking, it is less clear who is responsible. Under the Advocates Act 1961, a lawyer stays responsible for the work they file or adopt, and professional misconduct is punishable under section 35. The Supreme Court has now said clearly that citing an unverified AI-generated precedent is misconduct. A lawyer cannot blame the software. Once the lawyer adopts the output, it becomes the lawyer’s own.

E. Copyright and Ownership

Who owns what an AI produces? The Copyright Act 1957 does not give a clear answer. Section 2(d)(vi) says the person who causes a computer-generated work to be created is its author. That wording was added in 1994, long before generative AI. It is not clear whether it covers content that an AI system produces mostly on its own. The Department for Promotion of Industry and Internal Trade published a working paper on generative AI and copyright in December 2025. But a working paper is not law.

Law firms face practical questions. Who owns a draft that AI helped to write? Must the firm tell clients or courts that AI was used? Parliament or the courts will have to answer.

F. Cybersecurity

Law firms hold sensitive information, and hackers know this. Every outside AI service that receives client material is one more way in for an attacker. Many smaller Indian firms do not have the technical help to judge how safe a cloud-based AI tool is. The profession has not yet dealt with this in an organised way.

VI. The Indian Legal Framework

A. The Information Technology Act 2000

The Information Technology Act 2000 covers electronic transactions, data security and cyber offences. It does not mention AI. But its rules on unauthorised access and data security would apply if an AI platform were breached. Until the main parts of the Digital Personal Data Protection Act begin to apply, section 43A and the 2011 data security rules stay in force. They require reasonable security practices for sensitive personal data.

B. The Digital Personal Data Protection Act 2023

The Digital Personal Data Protection Act 2023 will change how lawyers use AI tools with client data. The Rules were notified in November 2025 and are being brought in step by step. The Data Protection Board came first, and most duties for businesses begin on 13 May 2027. A law firm that decides how and why client data is used is likely to be a “data fiduciary”. It must follow rules on consent, limited use of data and data security. It may hire a “data processor”, such as an AI vendor, only under a valid contract, and it stays responsible for compliance either way. The Act also has exemptions, for example for enforcing legal claims.

How all this fits into engagement letters, client consent forms and vendor contracts is something the profession has not yet worked out.

C. Government Policy on AI

NITI Aayog’s 2018 National Strategy for Artificial Intelligence called AI central to India’s growth and asked for responsible use. In November 2025 the Ministry of Electronics and Information Technology issued the India AI Governance Guidelines. They are voluntary. They say India will depend mainly on existing laws, such as the IT Act and the Data Protection Act, and will not pass a separate AI law for now. Both are general policy documents. Neither sets rules for the legal profession.

D. Courts and the Bar Council

Courts have started to act. In July 2025 the Kerala High Court issued a policy for its district judiciary. Judges may not use AI tools to reach findings, grant relief or decide cases. Only approved tools are allowed, and every use of AI must be recorded.

The biggest step came in July 2026 in Pooja Ramesh Singh v Jammu and Kashmir Bank Ltd. The NCLT had relied on six precedents that were fake, wrongly cited, or quoted paragraphs that do not exist. The Supreme Court set aside the tribunal orders. It called for zero tolerance of unverified AI-generated precedents and held that citing them is misconduct by an advocate. The Court made clear that it was not banning AI. Its worry was fake material being passed off as real authority. It also asked the Bar Council of India to set up a committee to frame guiding principles and disciplinary consequences (n 2).

The ruling came a month after the Supreme Court’s AI Committee, chaired by Justice P S Narasimha, released draft Regulations for Use of Artificial Intelligence in Courts, 2026 for public comment. The draft would apply to all courts and tribunals. It proposes a ban on using AI to decide cases. It would also require lawyers and litigants to tell the court when they have used AI to prepare filings. Comments closed on 20 June 2026, and at the time of writing the regulations were still a draft.

The Bar Council of India has also acted. Its circular of 17 July 2026 sets digital ethics rules for advocates, law students and interns. It warns against deepfakes, fabricated judgments and misleading AI-generated legal content. New applicants for enrolment must sign an undertaking on responsible digital conduct. This is still not the full set of rules on using AI in legal work that the Supreme Court has asked the Council to prepare. Meanwhile, the Supreme Court’s e-Committee continues to push digital court records and virtual hearings under Phase III of the eCourts project.

VII. Global Developments

A. The European Union

The EU Artificial Intelligence Act, adopted in June 2024, is the first broad law in the world on AI systems. It sorts AI uses by risk. AI that courts use to help research and interpret facts and law is high-risk. It must meet strict requirements on transparency, human oversight, accuracy and bias testing. These rules were meant to apply from August 2026, but a 2026 amending regulation moved the date to 2 December 2027 for this category. The Act directly binds only the EU. Still, its effect is likely to go further, because global technology companies and international law firms will shape their own AI policies around it.

B. The United States

After Mata v Avianca, American judges began to issue standing orders on AI. Judge Brantley Starr of the Northern District of Texas was the first, in May 2023. He required lawyers to certify that any AI-generated text in their filings had been checked. In July 2024 the American Bar Association issued Formal Opinion 512. It applies the existing rules on competence, confidentiality, communication, honesty to the court, supervision and fees to generative AI. The opinion is persuasive but not binding.

The speed is what stands out. Within weeks of a widely reported incident, individual courts had set clear duties for lawyers. Indian courts are now starting to do something similar.

C. China

China’s Interim Measures for the Management of Generative AI Services took effect on 15 August 2023. Providers of generative AI services offered to the public must follow content rules and work to improve accuracy and transparency. In some cases they must also complete security assessments and filings. For its own courts, the Supreme People’s Court issued opinions in December 2022. They encourage wider use of AI in judicial work but keep it in a supporting role, with human judges making the decisions. This approach shows close state direction of AI in courts. It is very different from the case-by-case style of common law countries such as the United States.

VIII. Advantages and Disadvantages

AI is not all good or all bad. A fair assessment has to look at both sides.

A. The Case in Favour

The gains in efficiency are real. Research that took days can now be done, at least roughly, in hours. Lower costs can bring legal help within reach of individuals and small businesses who could not afford it before. AI drafting of standard documents cuts down repetitive errors and makes documents more consistent. Chatbots can give basic guidance to people who would otherwise get none. And processing documents in bulk is a big help in large disputes and in clearing backlogs.

B. The Case Against

The risks are just as real. Hallucination, which means confidently producing false information, has already led to sanctions and findings of misconduct in several countries, including India. Uploading client material to outside AI services raises confidentiality questions that nobody has settled. Biased training data can build unfairness into legal outcomes. When AI makes a harmful error, it is often unclear who is responsible. And every outside connection adds to the cybersecurity risk.

Two longer-term worries also matter. Junior lawyers who depend on AI without building their own analytical skills may become weaker. And lawyers whose practice depends on high-volume routine work may lose that work to tools that do it much more cheaply.

IX. The Future of AI in Indian Law

A. Help for Court Administration

Indian subordinate courts alone hold close to five crore pending cases, and the High Courts add about 64 lakh more. AI could help by sorting and prioritising cases, preparing summaries for judges before hearings, finding relevant precedents and spotting errors in filings. None of this would replace the judge. The Supreme Court’s draft regulations point the same way, since they allow AI in support roles and keep decisions with judges (n 3). What is still missing is political will, funding and trained staff.

B. Indian Legal Technology

India has a growing legal-technology sector of its own. Companies such as CaseMine and SpotDraft build tools around Indian statutes, case law and practice. As these products improve and good technology reaches more lawyers, AI use is likely to spread across the profession. That includes sole practitioners and small regional firms that could never afford such tools before.

C. New Skills and Legal Education

The lawyers of tomorrow will need to understand both the law and what AI can and cannot do. They will need to know how to give clear instructions to AI tools, how to read results based on probability, and how to decide when AI should be used and when it should not. These skills do not replace basic legal training. They are added on top of it. Law schools will have to include them in their courses.

D. Rules to Come

Rules for AI in legal practice are coming to India. The Supreme Court has already told the Bar Council to frame them (n 2), and the draft court regulations show the likely direction (n 3). They may come through Bar Council guidance, amendments to the Advocates Act, court practice directions, or sector rules under the Data Protection Act. Lawyers who understand how the technology works, and where it fails, will be in a better position to help shape these rules than to simply react to them.

X. Conclusion

AI will not replace lawyers. It cannot sit with a grieving family through a succession dispute and offer real empathy. It cannot win the trust of a person facing criminal charges. It cannot exercise the ethical judgment that good lawyering needs, or answer an unexpected question from the bench with wisdom gained from experience. Those things will stay human for a long time.

What AI does change, and is already changing, is how legal work is divided. Tasks that fill a lawyer’s day without needing their full attention are being handed to machines. That puts judgment, empathy, strategy and ethical reasoning at the centre of the job. This does not make lawyers less valuable. The weight of their work shifts, and if the shift is handled well, clients should be better served.

The real danger lies elsewhere. A lawyer who avoids AI because it is unfamiliar may lose work to one who has learned to use it well. In a country with so much unmet legal need, the stakes are high for the profession and for society. The July 2026 ruling shows the other side: a lawyer who uses AI carelessly will pay for it.

The market has already decided whether lawyers will use AI. What remains is how. My answer is to use it carefully, check everything, and keep the discipline the law has always expected from its practitioners.

About Author

Bhagvati Vasava is a student at The Maharaja Sayajirao University of Baroda, Gujarat, Faculty of Law, and a dedicated legal researcher and academic with experience working with both international and national research journals, as well as reputed think tanks, including the International Institute of SDGs and Policy Research. She has completed internships across all levels of the judicial hierarchy, including District Courts, High Courts, and the Supreme Court of India, gaining comprehensive practical exposure to the legal system. She holds a strong command over legal academics and research methodology and actively engages in scholarly writing. Her research focuses on critical human rights and contemporary legal issues such as honour killing and ecocide. Through rigorous research and dedication, her work has been published in international-level journals, contributing to legal awareness and policy-oriented discourse and Committed to continuing her journey in the legal field, she strives to use research, writing, and advocacy as tools to promote justice, awareness, and progressive legal reform.

Bibliography

Legislation and Rules (India)

Advocates Act 1961

Bar Council of India Rules, Part VI, Chapter II (Standards of Professional Conduct and Etiquette)

Bar Council of India Rules of Legal Education 2008

Bharatiya Sakshya Adhiniyam 2023

Copyright Act 1957

Digital Personal Data Protection Act 2023

Digital Personal Data Protection Rules 2025

Information Technology Act 2000

Information Technology (Reasonable Security Practices and Procedures and Sensitive Personal Data or Information) Rules 2011