Legal advice has always carried consequences. A wrong opinion on a limitation period costs a client their cause of action. An incorrect assessment of a regulatory obligation exposes a company to enforcement proceedings. An erroneous reading of a will can disinherit a family. The legal profession carries these consequences through a combination of professional liability under the Advocates Act, 1961, disciplinary oversight by the Bar Council, and the implicit contractual relationship between lawyer and client. These are imperfect mechanisms, but they exist. When AI-generated legal advice causes the same categories of harm, none of these mechanisms apply, and Indian law currently has nothing to put in their place.
The deployment of AI in legal services is not a speculative concern. Large language model-based tools are already being used by law firms, legal technology companies, and individual practitioners for contract review, legal research, document drafting, and client-facing query resolution. Products such as law firm chatbots, AI-powered legal research engines, and consumer-facing apps that answer legal questions with apparent specificity are available to Indian users. The Bar Council of India has not issued any formal guidance on the use of AI tools in legal practice, and the Advocates Act, 1961 does not address the use of technological tools in the provision of legal services at all.
The liability gap operates at multiple levels. First, there is the question of professional liability. When an AI tool produces incorrect legal advice that a lawyer adopts and forwards to a client, the lawyer remains liable, but the degree to which the AI tool's developer shares that liability is entirely unresolved in Indian law. The Information Technology Act, 2000 (ITA) does not create a product liability framework for AI outputs. Section 79 of the ITA, which provides safe harbour for intermediaries, was designed for content moderation and platform liability, not for the liability of AI systems that generate affirmative professional advice. Applying Section 79 to an AI legal advice tool would require a significant extension of its logic, and even if such extension were accepted, it would protect the platform, not create a remedy for the harmed user.
Second, there is the question of consumer liability. The Consumer Protection Act, 2019 defines "deficiency in service" broadly enough to potentially cover incorrect professional advice provided through a digital platform. A consumer who pays for a subscription to an AI legal advice tool and receives advice that causes financial or legal harm could, in theory, bring a complaint before a Consumer Disputes Redressal Commission. However, AI legal advice tools typically include extensive disclaimer clauses stating that their output does not constitute legal advice and should not be relied upon without independent professional verification. The enforceability of such disclaimers against unsophisticated users, who may not understand the distinction between legal information and legal advice, has never been tested in Indian courts. In a consumer context, blanket disclaimers that operate to deny liability for a service's core function may constitute an unfair term under Section 2(46) of the Consumer Protection Act.
Third, and most structurally, there is the question of unauthorised legal practice. The Advocates Act, 1961 restricts the practice of law to enrolled advocates. Section 29 provides that only advocates are entitled to practise the profession of law. "Practising law" has been interpreted to include giving legal advice in a professional capacity. An AI tool that provides specific legal advice, telling a user that their employer has violated the Industrial Disputes Act, or that their landlord cannot legally terminate their tenancy, is arguably engaging in the practice of law. The developer of such a tool is arguably enabling unauthorised legal practice. This is not a reductio ad absurdum, it is a straightforward application of the statute's text to a novel technological context.
The regulatory responses available to India draw on international experience. The Law Society of England and Wales and the American Bar Association have both published guidance on the ethical use of AI in legal practice, including obligations of competence, disclosure, and supervisory responsibility when AI tools are used. The EU's AI Act, which came into force in 2024, classifies certain AI systems used in the administration of justice as "high-risk" systems subject to mandatory conformity assessments, transparency requirements, and human oversight obligations. India's National Strategy for Artificial Intelligence acknowledges AI regulation as a governance priority but has not produced binding rules.
What is required in India is, at minimum, a regulatory notification under the ITA requiring AI legal tools to register with a designated authority, maintain audit logs of outputs, disclose their AI nature to users before delivery of advice, and carry mandatory professional indemnity insurance. Beyond this threshold, a risk-tiered AI liability framework, as recommended by several legal scholars, would create proportionate obligations based on the potential for harm. Legal advice is a high-stakes domain. The cost of wrong advice is not a minor inconvenience; it can mean a missed limitation period, a forfeited right, or a prison cell. The law cannot afford to be as uncertain about AI liability as AI itself is about the law.