Singapore's MAS to Train AI on Live Bank Data in Anti-Financial Crime Push

_{Photo credit: Zhu Hongzhi / Unsplash.com}

Singapore’s central bank, MAS, is moving to deepen its use of artificial intelligence in financial crime prevention, launching a new industry initiative that will use live bank account data to train more accurate detection models.

The Monetary Authority of Singapore is conducting a proof-of-value exercise with five major banks to explore how AI and machine learning can improve pre-emptive scam detection across the financial sector. By pooling transaction data and account information, the project aims to identify higher-risk transactions and suspicious accounts earlier, enabling faster intervention before losses occur.

According to the regulator, historical transaction data from participating institutions will be used to train and evaluate the models in a secure data-sharing environment. The initiative also involves the Government Technology Agency and the Singapore Police Force, reflecting a broader public-private push to strengthen anti-scam defenses.

MAS said the data used in the exercise will remain confidential and protected by cryptographic safeguards, while a governance framework has been put in place to regulate how customer information is shared and used. The authority described the project as groundwork for deeper industry collaboration and wider deployment of AI tools against financial crime.

Financial institutions globally increasingly look to AI to improve fraud detection and transaction monitoring, but Singapore’s approach stands out for testing industry-wide models trained on shared, live banking data rather than siloed institutional datasets.