India’s markets regulator is weighing a domestically governed artificial intelligence model built specifically for the securities market, as it adopts a global supervisory toolkit for AI. Securities and Exchange Board of India chairman Tuhin Kanta Pandey said on October 28, 2026, at a summit in Mumbai, that SEBI is implementing the International Organization of Securities Commissions’ toolkit to strengthen risk management and build an agile governance framework for AI use in markets.
What SEBI Is Considering
The proposal under exploration is a securities-market-specific AI model that could be built and governed domestically, as a form of digital public infrastructure. That framing matters. Rather than simply buying or licensing a general-purpose model, SEBI is considering an AI system shaped by, and accountable within, India’s own market oversight regime. Pandey did not announce a build decision, a timeline, or a technical design; the model is at the consideration stage, alongside the toolkit implementation.
The regulator’s reasoning, as set out at the summit, starts with what AI can already do for markets. AI-enabled tools could improve services, risk assessment, and supervision, Pandey said. A model trained on market data could help spot anomalies, assess risk faster, and support supervisors sifting through vast volumes of trading and disclosure information. But he paired that promise with three conditions: reliable data, effective oversight of models, and clear accountability for decisions. Those conditions are the core of the governance problem, and they explain why a regulator would want a model it can govern rather than one it merely uses.
The IOSCO Toolkit And Why It Matters
The International Organization of Securities Commissions, the global body of securities regulators, has developed a supervisory toolkit for AI that member regulators can adapt to their own markets. SEBI’s adoption of that toolkit signals alignment with international peers at a moment when every major market is wrestling with the same questions: how to supervise AI-driven trading, advice, and risk systems without stifling innovation or creating new systemic risks.
A shared toolkit does not mean identical rules. Pandey described an agile governance framework, suggesting SEBI wants rules that can evolve as AI capabilities change, rather than fixed requirements that go stale within a year.
Data, Oversight, And Accountability
The three conditions Pandey named deserve unpacking, because they are where AI governance usually fails. Reliable data is the foundation: a model trained on incomplete, biased, or manipulated market data will produce confident but wrong assessments, and in securities markets wrong assessments can move prices. Effective oversight means the regulator, and the firms deploying models, can understand, test, and challenge what a model does, rather than treating it as a black box. Clear accountability answers the question that follows every AI error: who is responsible when a model-driven decision harms investors or destabilises a market?
A domestically governed model could, in principle, help on all three fronts. Data governance could be set under Indian law and market rules, oversight could be built into the model’s design and testing, and accountability could sit with identifiable institutions rather than distant vendors. But it would also concentrate risk: a single public-infrastructure model becomes a single point of failure, and its governance would need to be genuinely independent and technically competent to earn market trust.
The Wider Reform Agenda
Pandey also used the Mumbai summit to discuss broader market reforms. SEBI is reviewing compliance requirements for foreign investors and will issue a consultation paper seeking views on aligning them with global standards. While India has streamlined market access, eased digital onboarding, and simplified procedures for overseas investors, some domestic compliance requirements remain more onerous than those in other major jurisdictions, he said.
“India will need both domestic and foreign capital to support its growth ambitions,” Pandey said. “International participation can widen the pool of capital and bring expertise.” The AI discussion sits inside that larger ambition: a market that wants deeper foreign participation needs supervision that foreign investors trust, and AI is becoming part of how that supervision works. The Reuters report also noted that SEBI planned in August to relax disclosure requirements for certain offshore funds deemed high-risk, after rules introduced during scrutiny of offshore investors linked to the Adani Group triggered complaints and legal challenges.
Why It Matters
This matters because securities regulators worldwide are moving from warning about AI to building with it, and the choice between buying a model and governing one will shape market oversight for years. A domestic model could give India more control over data, testing, and accountability, but it also raises hard questions about cost, capability, and independence that a consultation paper alone will not answer. The same tension between AI capability and control runs through Malaysia’s seizure of smuggled AI chips, bots flooding a UK public consultation, and Bank of America’s warning about sharing passwords with AI: powerful systems are entering regulated spaces faster than the rules for governing them. SEBI’s toolkit adoption is a step toward closing that gap, and its domestic-model idea is the more ambitious question behind it.
Background
The Securities and Exchange Board of India regulates India’s securities markets. On October 28, 2026, its chairman said at a Mumbai summit that SEBI is implementing the IOSCO supervisory toolkit for AI and exploring a domestically governed, securities-specific AI model as digital public infrastructure, while also reviewing foreign-investor compliance requirements.
Source: Reuters report
0 Comments