More than half of senior banking leaders across Asia Pacific are not confident their institutions can detect when a customer is being manipulated into making a scam payment, according to a new poll by FICO.

The survey of 51 senior executives and C-suite leaders from banks across the region, conducted during FICO’s Banking Leaders Forum in Bali in August, found that 51 per cent of respondents are not confident their bank can detect customer manipulation before a payment goes through — including 49 per cent who said they were “not very confident”.

Banks want to act earlier, but lack the tools

Despite the confidence gap, banking leaders showed strong appetite for early intervention. Seventy-seven per cent said banks should act immediately once behavioural warning signs point to a significant risk, rather than waiting for a suspicious transaction to be attempted. Only 8 per cent said they would wait until a transaction is actually flagged.

The gap between ambition and capability is stark: 54 per cent identified stronger behavioural analytics and connected fraud intelligence as the top capability banks need to spot changes in customer behaviour early. Thirty-nine per cent pointed to the inability to share scam intelligence quickly enough across banks, telecommunications providers and social media platforms as a major weakness.

“Scammers are increasingly manipulating legitimate customers into authorising payments themselves, rather than trying to defeat the bank’s security controls directly. As AI makes social engineering more convincing and personalised, banks need to recognise changes in customer behaviour and intervene before the money leaves the account,” said Dattu Kompella, Managing Director in Asia Pacific for FICO.

AI-generated scams are the top emerging concern

Looking ahead, 46 per cent of respondents named AI-generated, personalised scam messages as their biggest emerging threat over the next three years, ahead of scam-as-a-service platforms (28 per cent) and deepfake voice and video impersonation (8 per cent). Fraud and scam prevention was also the most common priority banks cited for AI and agentic decisioning investment, chosen by 75 per cent of respondents.

Scaling that AI-driven decisioning remains a challenge in itself: 47 per cent of respondents cited data and infrastructure gaps as the biggest hurdle, while 43 per cent pointed to legacy systems and integration issues.