The Supreme Court wants AI to assist courts, but not replace judicial decision-making.
Draft regulations define where AI can be used and prohibit it from deciding cases, bail or witness credibility.
The move follows cases where AI-generated fake judgments found their way into judicial orders.
The Supreme Court has sent an unambiguous message on the future of artificial intelligence in India's justice system: AI can help courts work faster and more efficiently, but it cannot be allowed to influence judicial decision-making without strict safeguards. That message comes at a time when the judiciary is not only confronting the risks posed by generative AI but is also preparing a regulatory framework to govern its use.
The issue gained urgency after the Supreme Court set aside orders passed by the National Company Law Tribunal (NCLT) and the National Company Law Appellate Tribunal (NCLAT), finding that they relied on fabricated and inaccurate judicial precedents generated through AI. Around the same time, a Supreme Court committee released draft regulations that seek to define how AI can be used across courts, tribunals and other adjudicatory bodies within India's judicial system. Together, these developments underline a broader objective: embrace technology where it improves access to justice and court administration, but ensure that the responsibility for deciding disputes remains with judges.
Why the Supreme Court felt regulations were necessary
The immediate trigger came from an insolvency dispute involving Essel Infraprojects Ltd. As reported by The Indian Express, the NCLT relied on six judicial precedents while rejecting the company's objections to insolvency proceedings initiated by Jammu and Kashmir Bank.
When the case reached the Supreme Court, it found that three of the cited judgments did not exist. The remaining citations either attributed legal propositions that never appeared in the original judgments or referred to different cases altogether. Jammu and Kashmir Bank informed the court that it had not cited any of those judgments, indicating that the material appeared to have emerged during the tribunal's own research. The errors remained undetected even when the NCLAT upheld the tribunal's order.
For the Supreme Court, the implications extended well beyond a single insolvency dispute. If fabricated precedents could become part of judicial reasoning and survive scrutiny across multiple forums, the integrity of the adjudicatory process itself was at risk. The court therefore set aside the orders and directed that the matter be reconsidered.
The Bench also asked the Bar Council of India to constitute a committee to examine the problem of AI-generated legal authorities being cited without proper verification and to recommend safeguards for the legal profession. The direction recognises that maintaining the integrity of legal research is a shared responsibility of both the Bench and the Bar.
The judgment was not an isolated response. Earlier this year, the same Bench dealt with another case in which a trial court had relied on AI-generated, non-existent judgments. As The Indian Express reported, the recurrence of such incidents reinforced the need for clear standards before AI becomes more deeply integrated into legal practice.
Human primacy: The core principle
While the Supreme Court has acknowledged AI's potential to modernise court administration, it has consistently maintained that judicial authority cannot be delegated to technology.
That principle forms the foundation of the draft Regulations for Use of Artificial Intelligence (AI) in Courts, 2026. According to The Indian Express, the proposed framework states that AI systems should function only in an assistive capacity and must remain subordinate to human judgment. The authority to determine questions of law, assess facts, evaluate evidence and deliver justice will continue to vest exclusively in judicial officers.
This approach reflects a broader concern that efficiency should not come at the expense of independent reasoning. AI can process information at remarkable speed, but it cannot replace the exercise of discretion, accountability and constitutional responsibility expected of a judge.
What AI can—and cannot—do in courts
The draft regulations encourage the use of AI where it can simplify court administration and improve access to justice.
Permissible applications include case management, preparation of cause lists, hearing schedules, transcription of proceedings, translation of judgments, legal research assistance, record management and administrative support. AI-powered chatbots may help litigants understand court procedures, while accessibility tools can assist persons with disabilities and those facing language barriers.
The framework, however, draws firm boundaries around decision-making.
As reported by The Indian Express, AI systems would not be permitted to determine judicial outcomes, recommend sentences or independently influence adjudication. They would also be barred from assessing bail eligibility, predicting recidivism, evaluating flight risk or determining the credibility of parties and witnesses. The draft further prohibits the use of opaque or unexplainable AI systems in proceedings that may affect an individual's legal rights or personal liberty.
It also proposes safeguards relating to privacy and fairness by requiring AI systems to comply with the Digital Personal Data Protection Act, 2023, and to rely on lawfully obtained, representative and non-discriminatory data.
Global approaches to AI in the judiciary
India is not alone in confronting these questions. Courts across the world are experimenting with AI to improve efficiency while placing limits on its role in judicial decision-making.
Courts in the United States have reminded lawyers that AI-generated legal research must be independently verified after fabricated citations surfaced in litigation. Judicial guidance in the United Kingdom similarly permits the responsible use of generative AI but makes lawyers and judges accountable for checking accuracy. The European Union's AI Act adopts a broader risk-based framework for regulating AI across sectors, including certain high-risk public-sector applications that may affect individuals' rights.
India's proposed framework broadly aligns with this emerging international consensus: AI may support the administration of justice, but responsibility for judicial decisions cannot be transferred to machines.
Risks of AI hallucinations and bias
The Essel Infraprojects case illustrates one of the most significant limitations of generative AI—hallucinations, where AI systems produce information that appears convincing but is false.
As The Indian Express reported, the disputed citations included judgments that never existed alongside genuine decisions that were attributed with legal principles they never contained. If such material enters judicial reasoning unchecked, it can compromise the quality and credibility of legal decision-making.
The draft regulations also address algorithmic bias. They require AI systems deployed in courts to avoid discrimination on grounds such as caste, religion, gender, disability, language or economic status, while emphasising additional safeguards for vulnerable and marginalised communities. The objective is to ensure that technology enhances access to justice without reinforcing existing inequalities.
What the regulations could mean for India's legal system
The proposed framework goes beyond defining acceptable uses of AI. It also seeks to establish institutions that can oversee its responsible adoption.
The draft recommends creating a permanent apex body within the judiciary to frame standards, guide policy, oversee governance and encourage responsible innovation in the use of AI. It also proposes dedicated AI committees in the Supreme Court and every High Court to supervise implementation within their respective jurisdictions and ensure uniform standards.
If implemented, the framework could provide judges, lawyers and litigants with greater clarity on how AI may be used while preserving public confidence in the justice system. Rather than reacting to technological developments after problems arise, the judiciary is attempting to put institutional safeguards in place before AI becomes integral to everyday court functioning.
When will the rules take effect?
The proposed regulations remain in the consultation stage. The Supreme Court AI Committee has extended the deadline for stakeholders and the general public to submit comments on the draft regulations till July 15, 2026, after which the framework is expected to be finalised before being notified.
Although the final regulations are yet to be notified, the direction is already clear. India's judiciary is preparing for an era in which AI becomes an important part of court administration, but not of judicial authority. The Supreme Court's message is straightforward: technology may help deliver justice more efficiently, but the responsibility for delivering justice itself must remain with human judges.
(With inputs from Indian Express)
