Bank Negara Malaysia (BNM) and Payments Network Malaysia (PayNet) are building a sophisticated AI fraud detection system slated to go live next year.
The initiative is a direct technological counterpunch to the staggering financial losses from fraud, which reached a record RM1.58 billion in 2024 and continued with RM1.12 billion in losses reported just in the first half of 2025 .
The core of the new system is its ability to act as a digital sentinel, analysing transactions in the moments before they are completed. It will leverage machine learning, large language models, and predictive analytics to spot unusual activity, like a sudden large transfer to a suspicious account . When a risky transaction is identified, the system will instantly alert the user, prompting them with a confirmation like, “Are you sure you want to make this transfer?” This simple interruption is designed to break the scammer’s spell and give people a crucial chance to stop the transaction .
This project is part of a broader digitalisation push by BNM, which includes the recent launch of a Digital Asset Innovation Hub that provides a secure sandbox for testing new financial products . The central bank’s deputy governor, Datuk Marzunisham Omar, noted that this tech drive is happening alongside widespread AI exploration in the sector, with a survey revealing over 70% of Malaysian financial institutions are already implementing or looking into AI and generative AI tools .
In a parallel effort, PayNet is also developing a specialised, federated AI model tentatively dubbed FinancialGPT. This model is being trained entirely on local Malaysian financial data . This approach ensures the AI is highly attuned to local scam patterns while also addressing data sovereignty concerns by ensuring sensitive information never leaves the country.