The Securities and Exchange Board of India has flagged more than 20,000 instances of allegedly fraudulent or suspicious financial content on social media since November 2025, through an artificial intelligence surveillance system that arrives at a moment when the regulator’s own data shows 62 per cent of investors now shape their decisions around what financial influencers tell them online.
The system, called Project SUDARSAN, stands for Surveillance of Unauthorized Digital Activity via Real-time Scanner for Anti-fraud, according to SEBI’s Annual Report 2025-26. It continuously scans publicly available social media content, looking for material that could mislead investors, dispense unauthorised investment advice or impersonate regulated entities and legitimate institutions.
What the AI is built to catch
Project SUDARSAN is not confined to reading text. SEBI says the system can analyse videos, images, advertisements, messages and spoken content within videos, processing material across multiple languages including regional ones while attempting to assess context and likely intent. That range matters in a market where financial content increasingly circulates as short videos and voice notes rather than written posts.
The platform specifically hunts for claims of guaranteed or assured returns, fake certifications, impersonation of regulated entities, investment advice given without proper registration and unauthorised financial promotions. It combines these signals with behavioural patterns and regulatory parameters to generate risk scores, producing structured alerts that SEBI officials then review manually before any enforcement step is taken. SEBI has paired this with a second AI system, R(AI)DAR, aimed specifically at scrutinising financial advertisements for misleading or unauthorised promotional claims.
Crucially, identification by the AI does not by itself establish a violation. Flagged content still requires human and regulatory review, a distinction SEBI has been careful to preserve as it scales up automated monitoring alongside its conventional, manual regulatory processes.
Why the finfluencer economy worries the regulator
The scale of the surveillance effort reflects how deeply financial influencers have embedded themselves in ordinary investors’ decision-making. SEBI’s Investor Survey 2025 found that 62 per cent of investors base at least some investment decisions on finfluencer recommendations, and 93 per cent of respondents rated such influencers as moderately to highly credible. YouTube emerged as the dominant platform for investment-related information, cited by 91 per cent of respondents, ahead of Instagram at 64 per cent and Facebook at 61 per cent.
That trust has coincided with a string of high-profile enforcement actions that illustrate what SEBI considers the outer edge of permissible financial content. In December 2025, the regulator issued an interim order against trading educator Avadhut Sathe, impounding more than ₹546 crore in what it described as unregistered investment advisory services operating under the guise of paid education. The Securities Appellate Tribunal later reduced the deposit requirement on appeal, a modification the Supreme Court declined to disturb, indicating that even SEBI’s most aggressive interim actions remain subject to real judicial scrutiny.
A January 2025 circular had already sought to close a related loophole, barring individuals engaged purely in financial education from referencing security prices, including through code names or recent market data, and prohibiting SEBI-registered intermediaries from collaborating with unregistered influencers altogether. A follow-up circular later that year gave every regulated entity three months to terminate existing arrangements of that kind.
Technology as a supplement, not a substitute
SEBI has framed Project SUDARSAN as one layer within a broader shift toward technology-driven supervision, alongside measures such as validated UPI handles, the SEBI Check verification tool and verified app labels designed to help investors distinguish legitimate intermediaries from fraudulent ones. The regulator has also sought expanded legal authority to access encrypted communications on platforms such as WhatsApp and Telegram, reflecting how much of the unregistered advisory ecosystem has migrated into private groups beyond the reach of public-content scanning alone.
Even so, the sheer volume of flagged content, more than 20,000 instances in under a year, underscores the scale of the enforcement challenge SEBI faces relative to its capacity for manual review. Automated detection can surface suspicious material far faster than human surveillance teams ever could, but converting a flagged post into a substantiated enforcement action still depends on the same investigative and legal processes that predate the AI system.
For retail investors, the practical implication is that the presence of a large following, polished production values or apparent market knowledge offers no assurance of legitimacy. SEBI’s public guidance remains unchanged in substance even as its detection tools evolve: verify a person’s registration status independently, treat guaranteed-return claims as an automatic warning sign, and be sceptical of curated screenshots showing selective trading success. As financial content continues shifting toward video and vernacular formats that traditional oversight was never built to monitor, SEBI’s bet is that AI surveillance, layered onto its existing regulatory architecture, offers the clearest path to keeping pace.
