More than half of senior banking leaders across Asia-Pacific say they are not confident their banks can detect when customers are being manipulated into authorizing scam payments before the money leaves their accounts.
A survey of 51 banking executives conducted in August found that 51% lacked confidence in their institutions’ ability to identify customer manipulation before a payment is made. Only 2% said they were very confident, while 46% described themselves as somewhat confident.
The findings highlight a growing challenge for banks as scammers increasingly target customers themselves rather than attempting to break through conventional banking security systems. Criminals can manipulate legitimate account holders into making or approving transactions that appear authorized, making them more difficult to stop through traditional fraud controls.
Banking leaders overwhelmingly supported earlier intervention. About 77% said banks should act immediately when behavioral signals indicate a significant scam risk, while 10% favored warning customers and generally allowing them to decide whether to continue. Only 8% said intervention should wait until a suspicious transaction is attempted.
The biggest obstacle is identifying those warning signs early enough. Some 54% of respondents pointed to stronger behavioral analytics and connected fraud intelligence as the most important capability for detecting changes in customer behavior and improving scam prevention.
The problem extends beyond banks. About 39% of executives identified slow intelligence sharing among banks, telecommunications companies and social media platforms as a major weakness. A scam can begin on a social platform, move through communications networks and eventually result in a bank transfer, with critical warning signals potentially remaining fragmented between organizations.
Artificial intelligence is adding another layer of concern. About 46% of respondents identified AI-generated personalized scam messages and conversations as the emerging threat that worries them most over the next three years. Other concerns included scam-as-a-service operations, mule-account networks, deepfake impersonation and attacks targeting biometric authentication.
At the same time, banks are looking to AI and automated decision-making as part of the solution. Fraud and scam prevention was the most frequently identified priority for AI or agentic decision-making, cited by 75% of respondents. But implementation remains difficult, with 47% citing data and infrastructure limitations and 43% pointing to legacy systems and integration challenges.
The findings come as financial institutions globally confront increasingly sophisticated AI-enabled fraud. Banks have also warned that emerging AI agents used for online shopping could create new risks involving payment security, sensitive financial information and scams as automated systems become more involved in transactions.
The survey suggests that detecting fraud after a suspicious payment has already been initiated may no longer be enough. Banks are increasingly looking at behavioral changes, transaction patterns and signals from outside the financial system to identify possible manipulation before funds are transferred.
As scams become more personalized and harder to distinguish from legitimate activity, the ability of banks to connect intelligence, modernize fraud systems and intervene before a customer authorizes a payment could become a central part of financial security across the region.
