Conference programme

For all those who believe artificial intelligence is the answer, as recently as July 2026, the Supreme Court of India had to strike down decisions from AI researched judgments. This prompted the question of whether AI, like computers, lacks the ‘common sense’ to avoid generating results that solely please its programmers, rather than satisfying the rules of programming. We know computers cannot hallucinate; if large language models are fed with the wrong input questions, they provide wrong answers while ‘stealing’ data and dicta from vast libraries of information. The easy use of AI brings pop-up warnings that users dismiss as glitches of a new-age product but what happens when we want AI to help with constitutional morality, speeding up justice and maybe even justice delivery? How can AI speed up a legal system based on a 118-year-old Code of Civil Procedure and a Constitution that was created before the age of computers?

This opening plenary session will bring together senior judiciary, government and members of the Bar to confront questions at the heart of India’s common rule-of-law debate: how should constitutional morality guide the balance between judicial independence and institutional accountability? Can justice keep up with the demands of a rapidly modernising economy, measured by case pendency, enforcement timelines and greater access to justice? What does this mean for India’s credibility as a destination for international investment and dispute resolution?

SpaceX made history with an $86bn IPO, nearly double the GDP of South Africa where its CEO, Elon Musk, was born. Big money creates big disputes, raising ethical questions about justice delivery, constitutional principles and whether our judiciary can be truly independent. India is the world’s fastest-growing major economy, with recent IPOs in India showing unprecedented growth. But its legal infrastructure, such as the Indian Contract Act (1872), is mostly built on statutes drafted for a different era. This law now governs commercial relationships, technology platforms and cross-border transactions. The framers of these laws could not have anticipated today’s technological advancements, how agreements are made or even how quickly some businesses need disputes resolved. Insolvency reform took decades, data protection took longer and AI regulation is still incomplete. Contract enforcement remains slow, regulatory frameworks cannot develop even as their implementation is debated, the Courts are asked to resolve disputes arising from business models that the law has yet to name. This session asks the fundamental question that sits beneath every other discussion at this Symposium: is India’s legal system — its courts, its statutes and its regulatory architecture — keeping pace with the economy it is meant to serve, or is the gap between legal capacity and economic ambition becoming a constraint on growth?

There are no standard agreements for large projects and lawyers thus opt for an arbitral institution that’s familiar and convenient, but shouldn’t their advantages and disadvantages be weighed objectively? Which institution suits India best and which is most efficient? Senior arbitration practitioners will discuss their experiences, rule-related pitfalls and real-case anecdotes with some representatives from arbitral institutions who may decide to respond to anecdotal and objective criticisms which may emerge purely from this discussion.

Can computers really hallucinate? What is hallucination? How does it happen? How do computers furnish inaccurate results and how can this be avoided?

The panellists will discuss the Supreme Court of India’s recent regulations for use of artificial intelligence in courts and the session will also cover:

• The AI advantage to win your case – how does it work?
• Does AI really hallucinate or is this a misnomer for another real issue?
• Can we train lawyers to exterminate hallucination?
• What is AI capable of today and what are the relative advantages of different platforms
• Facing-off technologies (Harvey v Claude v ChatGPT v Legora, etc.)
• How to use AI for arbitration and drafting

With companies increasingly relying on self-learning models trained on similar public data sets, tacit collusion between competitors is a real risk. This panel will examine the conceptual and evidentiary challenges of proving algorithmic collusion, the adequacy of existing legal tools and whether new regulatory frameworks are needed.

There are two types of awards in international commercial arbitrations (ICA) under the Arbitration and Conciliation Act (1996): awards passed in India-seated ICA and awards passed in foreign-seated ICA.

In both cases, the execution of the award may be delayed. In the case of awards passed in ICA seated in India, delay may result on account of a challenge under section 34, proceedings for stay of the award (section 36), attachment of assets (section 9), appellate proceedings (section 37) or ordinary civil procedure. In the case of awards passed in ICA with a foreign seat, delay may result at the stage of execution when the execution is resisted, either under section 48 or section 57 of the Act. High Court judges, eminent senior counsel and a panel of well-known senior lawyers will examine these issues arising from post-award challenge and enforcement, in the case of awards passed in ICA in the Indian context.

The statutes and rules governing ‘electronic records’ in India have struggled to keep pace with the latest technology they are meant to regulate, while artificial intelligence has also widened this critical gap. The new Indian Evidence Act, the Bharatiya Sakshya Adhiniyam (BSA) (2023), expands the definition of electronic records to capture AI-generated data, but the foundational question of what makes digital evidence trustworthy remains unresolved. Currently, courts are receiving evidence that has been processed, enhanced or partially generated by AI tools without a framework for assessing the integrity, provenance or reliability of the model that produced it. Section 63 certificates, designed for a world of static files and server logs, are now required to authenticate deepfakes, synthetic audio, AI-generated documents and algorithmically reconstructed timelines. 

This session will examine evidentiary fault lines that practitioners are already encountering: how to establish and challenge the authenticity of AI-generated or AI-processed evidence, what metadata analysis and forensic examination can and cannot offer, how the new IT Rules’ synthetic-content labelling requirements interact with section 63 and what judges and opposing counsel should actually be asking when a document, recording or dataset has passed through an AI system before reaching the courtroom. A practical session for litigators who can no longer assume that the evidence in front of them is what it appears to be.

The panel will explore cross-border commercial dispute resolution, including recovery of money or other relief where parties and assets may be in different jurisdictions. For example, can India and other jurisdictions now sustain Mareva injunctions? Each speaker will discuss these issues arising in their jurisdiction or in another jurisdiction where they has protected client interests. The moderator, a High Court judge, will offer cross-border perspectives based on cases decided under Indian law.

India is in the midst of its most ambitious decriminalisation exercise in history. The Jan Vishwas (Amendment of Provisions) Act (2026) decriminalises or rationalises over 717 offences across 79 Central Acts, replacing imprisonment with civil penalties, administrative adjudication and compounding across sectors from banking to intellectual property. The Finance Bill (2026) grades punishment for tax defaults by quantum rather than mere non-compliance, and in Sandesara-Sterling Biotech the Supreme Court has endorsed settlement-driven resolution of high-value economic offences.

Yet India still recognises neither deferred prosecution agreements (DPAs) nor non-prosecution agreements and plea bargaining remains barred for offences carrying more than seven years, for those affecting the socio-economic condition of the country and for offences against women and children. The case for a DPA regime nevertheless resurfaces, driven by the pressure behind the UK’s Crime and Courts Act (2013) and the US Department of Justice’s policy: corporate fraud prosecutions are slow and, evidentially, punishing and prosecuting a company can do more economic damage than the offence itself.

Where should the line fall between a technical default and a wilful economic offence? Does substituting monetary penalties for imprisonment simply make non-compliance a cost of doing business for those who can afford to pay? One side will argue that an enforcement architecture already overburdened — the Serious Fraud Investigations Office, the Securities and Exchange Board of India, the Enforcement Directorate under the Prevention of Money Laundering Act would recover state losses faster under a judicially supervised DPA regime. The other will argue that prosecutorial discretion is contested enough in white-collar enforcement, that judicial capacity to scrutinise negotiated settlements is unproven and that DPAs let large corporates buy an accountability individuals cannot. Audience voting is to be undertaken.