India's securities regulator deploys AI to curb market manipulation and fraud

Inside Sebi’s Mumbai complex, analysts rely on artificial-intelligence platforms to sift through massive volumes of market data, algorithmic identifiers and social-media chatter for signs of misconduct. An AI-generated alert last month highlighted irregular activity in a newly introduced closing-auction process, leading the regulator to act against a JPMorgan Chase & Co. unit and later lift a trading ban after confiscating alleged illicit profits.

The watchdog’s AI push also targets pump-and-dump schemes by cross-checking suspicious tips with trading patterns, and it now files about 7,000 removal requests each month to platforms such as X, Instagram and Telegram, removing over 100,000 videos. Since 2019 Sebi has built a data centre and launched its own AI tool, Sudarshan, while expanding its technical staff to more than 200 analysts. Experts warn of challenges like data poisoning and limited GPU supply, so human reviewers must still validate AI outputs. Recent cases involving Jane Street underscore the regulator’s urgency to strengthen automated surveillance as India’s derivatives market approaches $4 trillion in notional value.