The use of Artificial Intelligence (AI) is increasingly becoming common in the legal profession, with lawyers, law firms and even courts using it for research, drafting and administrative tasks. While AI promises greater efficiency, it has also sparked concerns about inaccurate or “hallucinated” outputs making their way into judicial proceedings.The debate has gained urgency after the Supreme Court recently cautioned against the use of AI-generated material in courts while setting aside orders of the National Company Law Tribunal (NCLT) and the National Company Law Appellate Tribunal (NCLAT), which had relied on fake, AI-generated judgments.The apex court has also directed the Bar Council of India (BCI) to set up a committee to examine the issue of advocates submitting hallucinated or fabricated AI-generated material before courts as though it were a valid legal precedent.The top court’s intervention has triggered a wider debate — not on whether AI should be used, but how far its role should extend. Against this backdrop, The Indian Express speaks to legal experts on whether AI can be safely integrated into judicial and legal work, where the boundaries should be drawn, and what role regulators should play in governing its use.Not blind reliance, but as an aidExperts underlined that the issue was not about keeping AI out of the legal field, but about using it simply as an aid. Advocate Arya Tripathy, partner at Cyril Amarchand Mangaldas, says the Supreme Court’s judgment serves as a “timely” reminder that AI must remain an assistive tool rather than a substitute for human intelligence.Also Read | ‘Catastrophic’: Supreme Court warns against unregulated AI use in rulings“A blind reliance on AI-generated output for legal work is problematic at various levels. Not only does it impact the client’s rights, but it also dilutes the trust and privilege reposed and showcases poor professional ethics,” she says, adding that the apex court’s intervention basically seeks ethical and responsible use of AI.Embracing tech, its problemsTripathy acknowledges that the problem of overreliance and delegation of critical legal work has led to a steady rise in the problem of fake citations and incorrect legal positions.Story continues below this adIn June, Justice Wasim Sadiq Nargal of the Jammu and Kashmir and Ladakh High Court cautioned judicial officers against relying on AI-generated legal research without independent verification, after discovering incorrect and untraceable case citations in a trial court order.On the other hand, courts themselves are steadily embracing the technology.Referring to the iconic “Tarikh pe Tarikh” dialogue from the film Damini, the Allahabad High Court issued a series of instructions to the Uttar Pradesh government and police authorities in May in a bid to streamline the system. One such directive was for the police to implement a speech-to-text AI module to record witness statements.Meanwhile, acknowledging “rapid technological advancement” and the rise of tech-savvy judicial officers, the Gujarat High Court introduced a policy to regulate the use of AI in its judicial and administrative functions in April. It allowed judges and court staff to use AI as a controlled administrative and research aid, such as for translating documents, checking grammar and managing cause lists, while mandating human oversight.Story continues below this adRegulation vs supervisionAdvocate Mathuvanthy Mathavan, a partner at Poovayya & Co, says the complete negation of AI is “unrealistic”. Referring to the Supreme Court’s observation, she says the focus must be on proper regulation and supervision, “The judgment embraces AI as a useful tool while firmly insisting on human control.”The BCI should organise interactive sessions and training programmes for advocates on responsible AI use while prescribing “clear disclosure requirements” for its use in legal practice, Mathavan says, highlighting that there must be disciplinary consequences for filing fabricated, hallucinated, or unverified material before judicial forums.Also Read | https://indianexpress.com/article/legal-news/supreme-court-bci-expert-committee-ai-fake-judgments-10675231/The mandatory disclosure of AI tools is consistent with the Supreme Court’s draft Regulations for Use of Artificial Intelligence in Courts, 2026, which mandate disclosure of the origin of generative AI content and verification through a designated mechanism before such content is produced before a court.“To be unable to spot the hallucinations and bias inherent in these instances simply may indicate the need for lawyers to be trained as ‘AI-enabled lawyers’,” Tripathy points out. She cautions that merely prescribing norms may fall short if the accountability and transparency matrix is not suitably upgraded and suggests referring to the recently released International Bar Association’s guidance for the use of AI in mediation.Story continues below this adPermissible use vs over-relianceTalking about drawing lines between the permissible use of AI in judicial work, advocate Akshat Pande, managing partner at Alpha Partners, says overreliance on anything except one’s own intelligence is harmful.The use of AI by non-lawyers to do legal work instead of seeking professional advice could be more dangerous than lawyers misusing AI, he says. Mandatory disclosures may also affect the client as AI-generated drafts may be looked down upon by courts. “It will be better if courts also use technology to catch incorrect data, drafting or fake case quotations by AI and reject the drafts on that basis.”Mathavan too warns that overreliance on AI risks turning a useful aid into a substitute for independent thinking. “Human judgment and verification must always remain at the core of legal work.”Question of accountabilityAdvocate Anshul Verma, partner at SKV Law Offices, says that what matters more is who remains accountable when AI is used, and that “the responsibility stays with the advocate or the judge no matter what tool sits between them and their work.”Story continues below this adAlso Read | Hallucinations in the Supreme Court: Poor drafting isn’t just an AI problemShe says that with the Supreme Court’s latest intervention, now there is a clear opening for the BCI to define what verification should look like in practice, and set out a guiding principle to prevent similar occurrences, and back it with real disciplinary consequences.Pande, however, feels it would be best if the BCI stayed away from this. “They generally have a hard time understanding most modern facets of the legal profession; with AI, God knows what they will end up doing.”As courts, lawyers and regulators grapple with AI’s growing influence, the challenge will be to harness its benefits without compromising the accuracy and accountability that lie at the heart of the justice system. Pande sums it up: “It is irrelevant whether the judiciary likes it or not; AI will be used, and it’s better to get used to it and create systems to catch its misuse than to ban it.”