After a year of enthusiastic adoption, India’s legal fraternity is learning that artificial intelligence can occasionally be a little too imaginative. Law firms that eagerly signed up with platforms like Harvey and Legora are now pulling back, chastened by courtroom embarrassment over AI “hallucinations” – fabricated citations that looked convincing until they weren’t. The Supreme Court has called it out bluntly: not a glitch, but misconduct.
Aprajita Rana, partner leading the TMT (technology, media, and telecom) practice at AZB & Partners, agrees. “In the litigation sector, we are increasingly seeing cases where judges are critical of the misuse of AI for legal research and written statements by lawyers,” she says. “Yet they’re open to AI for routine tasks like case summaries, translations and preparation of procedural documents to improve access to justice.”
Beyond the Hype
Currently, the use of AI in the Indian legal sector is largely driven by client expectations. “Many firms are opting for third-party AI tools, and we are seeing more instances of such tools being used for routine legal tasks like summarising, case law review, proof reading and even legal research,” Rana adds.
Legal firms are also investing in collating their in-house templates and advice so that they can rely on quality datasets that reflect the working way of such firms. That said, the jury is still out on how such AI use will impact client billing practices.
India stands out as a complex market – rich in legal precedents but lacking Bar Council rules on AI liability or its limits in advice. Lawyers and firms must craft their own risk plans, potentially exposing clients to uneven views on privilege and responsibility. Nikhil Narendran, TMT partner at Trilegal, puts it bluntly: “AI is a powerful tool when used right, but careless use reveals old flaws. Hallucinations aren’t new; they’re just more visible now.”
Verification Mandate
Experts agree: lawyers can’t blame the tool. They must verify everything against primary sources. Savvy ones boost productivity with it; others over-rely or ignore it. “The problem isn’t AI – it’s skipping disciplined methods,” Narendran says.
Rana notes that in-house lawyers often push AI first drafts on external counsel, but fixes for errors can take longer than starting fresh. She sees hope ahead: better India-specific datasets will help. Additionally, corporate familiarity with AI will pressure firms to adapt, with global firms already mandating it or tweaking payment models. “Attorney liability hasn’t changed, so billing will get more varied,” she predicts.
However, Ramesh Vaidyanathan, founding partner at BTG Advaiya, pushes back a bit: “Don’t ditch AI entirely – it’s game-changing for accessible, efficient justice in India. Hallucinations just remind us: always verify sources.”
The Supreme Court’s message is firm: use AI for research help, not as a human replacement. Verify every citation. Courts like Kerala are eyeing rules requiring AI disclosure and human checks – a zero-tolerance stance on risks to court sanctity.