KUALA LUMPUR: Bank Negara Malaysia (BNM) and Payments Network Malaysia Sdn Bhd (PayNet) are developing an artificial intelligence-based fraud detection system targeted for rollout next year.
BNM deputy governor Datuk Marzunisham Omar said the system is being designed using machine learning, large language models , and AI to create predictive analysis from existing transaction data.
“We could alert the customer whether a bank transfer is safe or potentially fraudulent, and it will be able to ask ‘Are you sure you want to make this transfer?’ before the transaction is confirmed.
“This predictive analysis is something we are working on, and I hope we will be able to roll it out sometime next year,“ he said in a panel session at the release of the October 2025 World Bank Malaysia Economic Monitor report today.
Marzunisham said the system forms part of the central bank’s broader measures to combat mule accounts and strengthen public protection against online scams.
Marzunisham said BNM is pushing ahead with broader digital initiatives.
He pointed to the Digital Asset Innovation Hub, which he said provides a secure platform for regulators, financial institutions and private sector players to test and develop new digital products.
“This hub is an important initiative to encourage innovation in a safe environment,“ he said.
Marzunisham noted that finance has been one of the fastest sectors to digitalise, particularly in banking, payments and financial services.
He cited a survey by the Asian Institute of Chartered Bankers (AICB), which found that more than 70% of financial institutions are already exploring various forms of AI and generative AI.
“The industry, through AICB, is also developing a framework to be shared among industry players to guide responsible adoption of AI,“ he said.
According to news reports, PayNet is exploring the development of a federated AI model, tentatively called FinancialGPT, to enhance fraud detection and strengthen security in digital payments.
PayNet chief of staff Dr Endry Lim Zhen Wan said the large transaction modelmwould be trained solely on Malaysia’s financial transaction data to enhance resilience against cyber threats.
“We believe this model would tackle fraud and scams, ensure broader financial inclusion and address compliance challenges specific to our market while securing sensitive data never leaves the country.
“It combines global best practices with national sovereignty, making it uniquely suited to Malaysia’s needs,“ he said.
The Home Ministry recently stated that Malaysia incurred RM1.12 billion in losses from online scams in the first half of 2025. This follows a record RM1.58 billion in reported losses in 2024.
