AI-Trained Payments Fraud Detection System Under Works, RBI Says

The Reserve Bank of India is developing an artificial intelligence-trained fraud detection system to alert bank account holders about potential risky transactions to curb fraud.

"It is in an experimental stage. We are making the software," Governor Sanjay Malhotra said in a press conference after the Monetary Policy Committee on Wednesday.

The RBI's under-development 'Digital Payments Intelligence System' will be run by an entity, which will collect data from various sources to trace mule accounts and flag risks to both customers and banks, Deputy Governor T Rabi Sankar explained.

Officials did not share additional details about the system.

Last month, the RBI proposed new norms that will allow digital payment providers to implement additional risk-based checks beyond the mandatory two-factor authentication for certain transactions. The new measures are designed to enhance security by applying extra scrutiny to transactions that have a high-risk perception, thereby reducing the potential for fraud.

The proposed norms are also intended to foster a more open and flexible digital payments ecosystem. They facilitate interoperability and open access to technology, which could lead to greater innovation and a more seamless experience for consumers and businesses alike. Further, the new regulations will clearly define the responsibilities of card issuers.

They will be mandated to validate the Additional Factor of Authentication or AFA in non-recurring, cross-border card-not-present or CNP transactions whenever a request is made by the overseas merchant or acquiring bank. This clarification aims to streamline the process for international payments and ensure a consistent level of security.

Data security and transaction frauds have been a key issue for the banking regulator as Indians have increasingly used digital platforms like UPI for everyday and large transactions. Such payment mechanisms can expose both customers and financial institutions to sophisticated risks.