When AI governance discussions cover India's mutual fund sector, they typically focus on AMCs — the fund houses making investment decisions. RTAs are the operational infrastructure that AMCs run on: transaction processing, investor registers, KYC, dividend distribution. CAMS and KFin together process the operational backend for essentially the entire Indian mutual fund industry.
The regulatory obligations are separate. An AMC's AI governance obligations (SEBI AMC AI Surveillance 2024) cover investment-side AI — market abuse detection, portfolio monitoring. The RTA's obligations cover operations-side AI — the KYC matching, SIP processing, fraud detection, and investor data management that supports 15+ crore investor accounts. One regulatory framework covers both layers only in principle; in practice, each entity type has its own compliance obligations under SEBI's master circular architecture.
Scale Creates Unique Governance Risk
RTAs process investor transactions in batch cycles that affect millions of accounts simultaneously. A poorly governed AI model in the NAV or SIP processing pipeline does not create one error — it creates a systematic error replicated across millions of investor accounts in the same processing cycle. This systemic scale risk is distinct from the individual transaction risk that AMC AI governance addresses. The accuracy validation, fail-safe, and human oversight requirements for RTA AI systems need to account for batch-scale error propagation, not just individual transaction error rates.
SEBI RTA AI Governance Compliance Checklist
- ☐ Chapter III-A inventory: all AI/ML tools used in RTA operations documented; sole liability for outputs and investor data confirmed; vendor contracts reviewed for liability consistency
- ☐ SEBI AI/ML reporting: all AI/ML systems reported to SEBI per intermediary AI reporting framework; reporting current
- ☐ KYC matching AI governance: accuracy benchmarking completed; false positive/negative rate documented; investor redress mechanism for AI-driven KYC rejections operational
- ☐ Transaction processing AI safeguards: fail-safe controls for AI-enhanced batch processing; human oversight for high-value or anomalous transaction flags before execution; audit trail for AI-driven processing decisions
- ☐ Fraud detection AI compliance: PMLA compliance for AI-based suspicious transaction flagging; FIU-IND reporting integration for AI-detected suspicious activity
- ☐ DPDP Act compliance: 15+ crore investor records subject to DPDP Act purpose limitation, data minimisation, retention policies; AI tool data processing consent mapped; third-party AI vendor data access governed under RTA's sole accountability
- ☐ Cybersecurity and CSCRF: AI systems in RTA operations comply with SEBI CSCRF obligations; AI vulnerability detection tools deployed per SEBI May 2026 advisory