Expert SpeakYoung Voices
Published on Jul 18, 2026
Artificial intelligence is transforming how Indian lawyers learn, work, and exercise professional judgement, demanding reforms in legal education, regulation, and access to justice
Artificial intelligence is reshaping justice systems and professions worldwide, and India is no exception. Over five crore cases are pending in India’s courts, more than 85 percent of them in the district and subordinate judiciary. AI is remaking the legal profession just as forcefully. For India’s over one million enrolled advocates, most outside the small tier of AI-equipped commercial firms, the question is no longer whether generative AI will be used but how unevenly its costs and benefits will be distributed.
Every profession that depends on judgement has built that judgement through apprenticeship. Lawyers did not learn to advise clients by being told how. They learned by drafting hundreds of contracts and summarising thousands of pages of case files, absorbing through repetition the instincts that separate competent practice from textbook knowledge. Generative AI now performs much of that repetition itself. The consequence is not that lawyers become redundant, but that the profession’s primary mechanism for producing professional judgement is beginning to change.
The Changing Economics of Apprenticeship
Legal training has long been subsidised by billing. Clients paid hourly rates for associates to research and draft, and that revenue funded years of supervised repetition. The Thomson Reuters 2025 Future of Professionals report found that 80 percent of respondents expect AI to change how firms price, staff and deliver work, even though roughly 90 percent of corporate legal spend on outside counsel still runs through hourly billing. As clients resist paying by the hour for work AI can automate, fixed-fee arrangements are gaining ground, narrowing the revenue that once funded junior lawyers’ training.
Smaller practitioners and legal aid organisations such as NALSA remain outside the AI ecosystem, widening an already significant access-to-justice gap.
When billing stops rewarding volume, firms stop generating volume for juniors to learn from. AZB & Partners rolled out Harvey AI firm-wide in September 2025 for document review and translation, following Shardul Amarchand Mangaldas’ earlier partnership with the platform. Harvey has since opened a Bengaluru office to serve a legal services market projected to reach US$67.4 billion by 2030. The International Bar Association has observed automation strikes first at tasks once assigned to junior lawyers, a pattern repeated in India. Hiring will likely shift toward fewer associates for drafting and more for supervising AI output from the outset of their careers. These gains concentrate among large commercial firms. Smaller practitioners and legal aid organisations such as NALSA remain outside the AI ecosystem, widening an already significant access-to-justice gap.
Competency Redefined
If junior lawyers no longer learn by doing the volume of work AI now performs, legal education must deliberately supply what apprenticeship once supplied by accident. Indian commentary on legal pedagogy has begun raising a related concern: when students let AI tools substitute for the cognitive effort of reasoning through a problem, they risk never developing the judgement that effort was meant to build. The Bar Council of India has directed universities to incorporate artificial intelligence into their curricula, but adding AI as a subject differs from training lawyers to supervise, interrogate and overrule it. An OECD analysis notes AI is more likely to reshape tasks than replace workers, particularly where it complements human skills. In law, this suggests AI’s value lies in supporting judgement, not replacing it, provided lawyers evaluate its output critically rather than accept it wholesale.
AI supervision is becoming a discrete professional competency, not a passive byproduct of new software.
The cost of treating AI supervision as incidental is already visible. In December 2024, the Income Tax Appellate Tribunal’s Bengaluru bench recalled its own order in the Buckeye Trust matter after discovering that judgments it had cited, apparently sourced through ChatGPT, did not exist. In September 2025, a Delhi High Court petition was withdrawn when opposing counsel proved the cited paragraphs, in one instance an entire judgment, were fabricated. By early 2026, the Supreme Court confronted more than a hundred imaginary citations filed in a commercial appeal. In July 2026, the Supreme Court set aside a tribunal ruling built on six fabricated precedents, ruled even an iota of hallucinated material voids a decision, and directed the Bar Council of India to frame disciplinary rules for unverified filings. Because every precedent’s authority must be independently verifiable, that burden falls on the lawyer, not the tool. AI supervision is becoming a discrete professional competency, not a passive byproduct of new software.
A Judiciary Setting the Pace
India’s courts have moved faster than the profession’s own regulators. In December 2025, Chief Justice Surya Kant reconstituted the Supreme Court’s Artificial Intelligence Committee. In June 2026, the committee published its most significant intervention yet, draft Regulations for Use of Artificial Intelligence in Courts, open for public comment until 20 June, later extended to 15 July. The draft requires lawyers and litigants to disclose AI use in pleadings, bars AI from deciding cases or assessing bail eligibility, and treats all AI output as advisory pending human verification, though disclosure alone does not test the output’s reliability. Adalat AI, a courtroom transcription tool, now operates in more than 4,000 courts, and, in July 2025, the KeralaHighCourt became the first Indian high court to issue a binding AI policy, barring its use in findings or judgments in the district judiciary. The government has earmarked 53.57 crore rupees under the eCourts Phase III programme for tools including LegRAA to support judges with legal research, and Digital Courts 2.1, a paperless case management platform, though government submissions to Parliament describe these tools as confined to a controlled pilot phase. Whether efficiency gains at the apex will ever reach the district and subordinate courts carrying most of the pendency burden, rather than merely modernising paperwork, remains an open question.
The Bar Council’s own Standards of Professional Conduct, framed before large language models existed, remain silent on AI delegation and verification duties. The American Bar Association’s Formal Opinion 512, issued in July 2024, established that lawyers using generative AI must discharge their duties of competence, confidentiality and supervision. India’s advocates have no equivalent, even as the judiciary regulates AI use more explicitly than the bar does. Confidentiality compounds the gap: AI systems often process client material on infrastructure outside India, raising questions the Digital Personal Data Protection Act, 2023 has not resolved for attorney-client privilege.
India’s Legal Profession Through the Next Decade
The contours of the next decade are already visible in these trends. Large firms will keep hiring junior associates, but fewer of them and for narrower purposes, as AI-supervision and legal-technology roles expand. Specialist litigation, regulatory advisory and courtroom advocacy, the work least reducible to pattern-matching, will retain its premium. As one Bar & Bench analysis of India’s legal AI market has observed, AI does not eliminate lawyers’ skills so much as it redefines which of them the market rewards. Smaller firms and solo practitioners, lacking the capital for enterprise AI tools, risk falling further behind even as litigants expect faster turnaround, widening the gap between large and small practices unless policy intervenes.
What Regulation Must Do Now
The most immediate priority is pendency. The Department of Justice and the eCourts Phase III programme should set a binding timeline for extending pilot tools such as LegRAA and Digital Courts 2.1 to the district and subordinate courts, where 85 percent of pending cases sit, with success measured by pendency reduction, not administrative efficiency.
The transition away from the apprenticeship model also requires institutional support. A government-backed shared AI infrastructure scheme, channelled through NALSA and state legal services authorities, would extend tools concentrated in large firms to legal aid providers and solo practitioners, keeping the access-to-justice gap from widening further.
Legal education has not kept pace with what supervision now demands. The Bar Council’s curriculum mandate should move beyond teaching AI as a subject and require competence in verifying, interrogating and overruling AI-generated work before enrolment.
The hallucination cases point to an immediate discipline gap. The Bar Council should adopt the disciplinary rules the Supreme Court has directed it to frame, making independent verification of every AI- authenticated domestic legal databases so verification does not depend on proprietary systems beyond domestic reach
Finally, confidentiality requires clearer safeguards. The Bar Council should mirror the Supreme Court’s disclosure duty in its own Standards of Professional Conduct, while the Ministry of Law and Justice reviews the Advocates Act and the Digital Personal Data Protection Act to close the confidentiality gap around attorney-client privilege.
The question is not whether lawyers will use AI, but whether India’s legal institutions will act on the gaps identified here — in verification, education and access — before the profession is reshaped by default rather than by design.
Conclusion
What is changing is not simply how lawyers work but how they become lawyers at all. The apprenticeship model that built professional judgement through volume is giving way to a profession where that volume is increasingly automated, with no institution yet replacing what apprenticeship taught. For India, this matters: a judiciary carrying tens of millions of pending cases cannot afford a profession that trains advocates too slowly, or an access-to-justice gap that widens as AI’s benefits concentrate among large firms. The question is not whether lawyers will use AI, but whether India’s legal institutions will act on the gaps identified here — in verification, education and access — before the profession is reshaped by default rather than by design.
Sairah Zahooris a Research Intern at the Observer Research Foundation.
Disclosure:Claude 4.6 was used for preliminary research support and language refinements.
The views expressed above belong to the author(s). ORF research and analyses now available on Telegram! Click here to access our curated content — blogs, longforms and interviews.
Author
Sairah Zahoor
Sairah Zahoor is a Research Intern at the Observer Research Foundation. …
Related Search Terms
Publications
When AI Becomes a Colleague: The Future of India’s Legal Profession
Artificial Intelligence
Jul 18, 2026
