An AI built at home
Pandey's most striking proposal is a securities market-specific AI model that would be built and governed in India. Framed as digital public infrastructure, the idea echoes the country's earlier successes with publicly owned digital platforms, such as the Aadhaar identity system and the UPI payments network, which became foundations on which private companies could build services.
A domestically governed model could, in principle, be trained on Indian market data, understand local regulations and languages, and be shared by brokers, exchanges and investors under common rules. It would also reduce dependence on foreign technology providers for a function as sensitive as market surveillance. Details on timing, cost and design have not yet been disclosed.
Governing a powerful technology
Alongside the model, SEBI is putting into practice IOSCO's supervisory toolkit for AI, aimed at strengthening risk management and building governance flexible enough to keep pace with a fast-moving technology. Earlier this year, Pandey said the regulator would issue guidelines on AI-driven trading and warned that the technology brings opportunities as well as heightened cyber risks, along with concerns over opacity, bias and the concentration of technological power.

















