NPCI-linked AI fraud pilots signal push for real-time payments risk controls
An Ebix executive highlights AI, secure APIs and real-time intelligence as fraud-prevention priorities, citing reported NPCI bank pilots of transaction-risk models and a proposed Digital Payments Intelligence Platform.
What happened
An Ebix executive argues Indian financial institutions should use AI, secure APIs and real-time risk intelligence to prevent fraud. The article cites NPCI’s
Key facts
- 6.58 million financial fraud complaints
- Rs 55,050 crore involved in reported fraud complaints
- 2021-2025
- Rs 220 million average cost of a data breach in India
- 13% year-on-year increase
Why this matters
Evaluate partnerships or tuck-in targets in transaction-risk models, secure payment APIs and fraud intelligence platforms that can strengthen real-time checkout protection in India.
What to watch
- Formal NPCI, RBI or bank announcements identifying live AI transaction-risk pilots, standards or rollout timelines.
- Launch details, governance structure or participation requirements for a Digital Payments Intelligence Platform.
- Gateway announcements offering network-level fraud scores, device fingerprinting or real-time merchant risk APIs.
- Changes in UPI payment failure rates, fraud complaints, chargeback-like dispute volumes or mandated step-up authentication.
- Evidence that fraud shifts from payment authorization to account takeover, refund abuse, mule-account activity or synthetic identities.
- Audit UPI and digital-wallet fraud exposure by payment method, device type, order value, geography and customer cohort.
- Ask PSPs, acquirers and payment gateways for roadmaps on real-time risk APIs, NPCI-aligned controls, false-positive rates and merchant override capabilities.
- Build checkout fallbacks for risk-triggered transactions, including alternate UPI handles, cards, pay-on-delivery where viable and low-friction reauthentication.
- Set fraud-control KPIs that jointly track chargebacks, account takeover, payment success rate, checkout abandonment and customer-support contacts.
- Review consent, data-retention and model-audit practices before sharing device, behavioral or transaction data with fraud vendors.
Also reported by
- YourStory · Capital — Same time