SEBI Sudarshan Explained India’s AI Watchdog for Market Manipulation
What Is Sudarshan?
Sudarshan is SEBI’s newly launched AI-powered market surveillance system, designed to:
• Detect fake financial analysts and misleading investment advice
• Enhance investor protection and market integrity
It’s part of SEBI’s broader push to integrate AI and machine learning into regulatory frameworks.
History and Background -
The Securities and Exchange Board of India (SEBI) has been progressively integrating artificial intelligence (AI) and machine learning (ML) into its regulatory framework to enhance market oversight, investor protection, and operational efficiency.
This evolution began gaining momentum in the early 2020s, with SEBI mandating regulated entities to disclose AI/ML usage as early as 2019.
By 2021-2022, SEBI invested in advanced data infrastructure, including data lakes for handling structured and unstructured data from over 550 crore daily order and trade messages in equity and derivatives segments.
These efforts focused on real-time surveillance to detect complex manipulations like front-running and insider trading.In June 2025, SEBI released a consultation paper outlining guidelines for the responsible use of AI/ML in securities markets, emphasizing ethical deployment, human oversight, and risk management across trading, compliance, surveillance, and advisory functions.
This paper proposed a five-point framework: model governance, transparency, fairness, robustness, and accountability. Sudarshan emerged as a key outcome of these initiatives, building on earlier AI tools like MITRA (launched in February 2025 for fraud detection via big data analytics).
Key Facts and Details-
Full Name and Purpose:
Sudarshan is SEBI's in-house AI-powered market surveillance system designed to monitor trading activities, detect anomalies, and counter misinformation in financial markets.
Scope:
It oversees real-time trading, processes IPO documents, analyzes market behavior, and scrutinizes investment advisors, research analysts, fake financial influencers, and misleading social media content.
Achievements:
Has flagged over 100,000 pieces of potentially misleading content, collaborating with platforms like Meta and Google for removals.
It speeds up intermediary approvals (e.g., mutual funds) and IPO scrutiny by efficiently identifying concerns.
Related Tools:
Complements R(AI)DAR (Regulatory AI-Driven Advertisement Reviewer), launched alongside for ad and social media monitoring, and earlier systems like MITRA.
Broader Impact:
Part of SEBI's push for a "lighter regulatory approach" for internal AI tools like surveillance, while stricter for client-facing models.
It aligns with the Cybersecurity and Cyber Resilience Framework (CSCRF) to combat ~70,000 fake UPI IDs impersonating SEBI entities.
How It Works -
Sudarshan leverages AI, machine learning, natural language processing (NLP), and big data analytics to process massive datasets in real-time.
Here's a step-by-step overview:
Data Ingestion:
Collects structured data (e.g., trading volumes, price patterns) and unstructured data (e.g., social media posts, advisor communications) from exchanges, brokers, and third-party sources like bank transactions.
Processes massive volumes of structured and unstructured trade data.
Pattern Recognition and Analysis:
Uses ML algorithms to identify anomalies, such as unusual trading volumes, wash sales, circular trading, or front-running.
NLP scans social media for misleading phrases, fake analyst claims, or illegal promotions.
Uses machine learning to detect anomalies, pump-and-dump schemes, and coordinated manipulation.
Flagging and Cross-Verification:
Flags suspicious activities by cross-referencing with off-market deals, historical patterns, and external data.
For instance, it links trading spikes to advisor posts to detect manipulation.
Flags misleading posts, fake finfluencer advice, and illegal promotions.
Alert and Action:
Generates alerts for human overseers, enabling swift interventions like trade halts or content takedowns. Predictive analytics forecast risks, reducing manual reviews.
Oversight and Iteration:
Incorporates human-in-the-loop governance for bias checks, model updates, and compliance with SEBI's ethical guidelines, ensuring transparency and fairness.
Timeline -
2019
SEBI mandates reporting of AI/ML applications by market intermediaries.
2021-2022
SEBI implements data lakes and AI for processing vast market data volumes.
February 2025
Launch of MITRA, an AI surveillance system for real-time fraud detection.
June 20, 2025
SEBI consultation paper on AI/ML guidelines released, seeking public feedback.
July 2025
SEBI enhances AI-driven surveillance for analyzing price patterns and trading volumes.
October 7-8, 2025
Sudarshan showcased at Global Fintech Fest; officially rolled out on October 8 as an AI-powered market surveillance system.
Ongoing (post-launch)
Integration with tools like R(AI)DAR for monitoring misleading ads and finfluencer
