From trades to balance sheets: Sebi to widen AI surveillance net

The market regulator plans to use a dedicated AI model to track quarterly results and flag financial misstatements and manipulation without waiting for investor complaints

The Securities and Exchange Board of India (Sebi) is planning to extend artificial intelligence (AI)-based surveillance from trading data to corporate filings, a dedicated AI model akin to its existing trading-surveillance tool. This would help flag financial misstatements and manipulation in quarterly results without waiting for investor complaints.

 

While the regulator’s internally developed AI models generate alerts based on which the bulk of enforcement action in trading-related violations are taken, corporate investigations remain largely complaint-driven.

 

The market watchdog has developed a team dedicated to fill this gap and work on the AI-model, Sebi wholetime member Kamlesh Chandra Varshney indicated at a recent conference on capital markets organised by the Federation of Indian Chambers of Commerce & Industry (Ficci).

 

With the tools, the regulator will be able to track quarterly results and identify violations such as manipulations in financial statements —without waiting for complaints, he added.

 

Regulatory experts said the Sebi’s pivot towards AI has helped in taking prompt action.

 

These plans on AI-based tighter surveillance and alerts build on tools Sebi has already deployed — from fast tracking approvals for initial public offerings (IPOs) and foreign portfolio investor (FPI) registrations to alerts on misleading social media content. They also include AI-driven review of advertisements of asset management companies (AMCs). 

 

The regulator also plans to extend its AI-driven review of advertisements beyond AMCs and expanding its outreach with Meta and Google to flag mis-selling and fraudulent content.

 

Apart from these tools, Sebi also adopted an in-house application called InfoMerge for the investigation department.

 

The portal automates multiple investigation activities, such as data acquisitions, analysis, and report generation to reduce the processing time for the case.

 

According to the regulator’s latest annual report, this system is being used to examine cases of violations of the Prohibition of Insider Trading (PIT) Regulations and Prohibition of fraudulent and unfair trade practices (PFUTP) Regulation.

 

During the year, it was enhanced to analyse KYC and trade-log datasets to spot suspicious patterns, generate AI-written summary reports, visualise suspect connections, and auto-generate summons and case notes — cutting processing time significantly.

 

Among other key technology adoptions was a Cloud-based platform for exchanging data and alerts on inspection with brokers and depository participants for faster corrective actions.

 

Last year, the market watchdog also deployed tools such as Regulatory (AI) Driven Advertisement Reviewer called R(AI)DAR and Project Sudarsan — both AI-powered tools.

 

While R(AI)DAR is utilised for reviewing advertisements for misleading claims, Sudarsan is for fraud-detection and scanning videos and content across social media. It has identified over 20,000 instances of fraudulent content in real time.

 

Further, the regulator also has an AI-enabled platform Cyber-Sec Audit Compliance (C-SAC) which automates cybersecurity audits of regulated entities. It had processed reports for eight market infrastructure institutions and 23 mutual funds in 2025-26. 

Game plan 

• New AI model to flag concerns in quarterly results, corporate filings
• Aims to detect fraud without waiting for investor complaints
• Dedicated team formed to build corporate-fraud AI model
• Other AI tools already in use include InfoMerge, Sudarsan, R(AI)DAR, and C-SAC
• AI-driven alerts already power bulk of trading violation enforcement actions