Banks cannot fight modern scams with fragmented data
They need connected information to tell a genuine transaction from a fraudulent one
THE recent joint operation between Singapore’s Anti-Scam Centre and five major banks, which prevented more than S$46 million in potential scam losses, demonstrates what can be achieved when information moves quickly between banks and law enforcement.
It also shows the uncomfortable reality that the nature of scams has changed more quickly than many of the controls designed to detect them.
More than 80 per cent of reported scams in the first half of 2026 involved victims transferring money themselves, rather than scammers taking direct control of their accounts.
For banks, this creates a fundamentally different challenge.
Traditional fraud controls are well-suited to spotting obvious anomalies such as an unfamiliar login, a stolen credential or a transaction that breaches a predetermined threshold.
But when a customer logs in normally, passes the required authentication checks and willingly authorises a transfer, the transaction can appear legitimate even when the customer is being manipulated by a scammer.
Detecting that kind of activity requires banks to look beyond the transaction itself and consider it in the context of what they already know about the customer.
A new beneficiary or unusual payment may take on greater significance when viewed in the context of changes in transaction behaviour or other indicators of risk.
Banks may already hold the information to make those connections, but more often than not, it sits across siloed systems and teams, or technology stacks that have been built independently of one another.
Know Your Customer (KYC) data can sit in one system, transaction monitoring in another, while sanctions screening and other compliance processes operate separately. As a result, a risk signal identified in one part of the bank might not make its way to another part of the bank.
Transaction monitoring could identify an unusual pattern of payments without that change being reflected in the customer’s broader risk profile, and information gathered through KYC may not be readily available to the team assessing suspicious activity.
Each system can be working as intended while the institution still lacks a complete view of what is happening.
Tackling this fragmentation is becoming increasingly important. KYC cannot be treated as a largely static exercise completed at onboarding and revisited periodically while transaction activity is assessed separately.
Connecting these parts of the compliance stack allows a change detected in one area to inform risk decisions elsewhere, giving teams a more current and complete view of the customer.
This is where artificial intelligence can help the most, when it is applied across connected customer and transaction data. It does this by identifying behavioural changes, drawing links between risk signals or surfacing unusual patterns that might otherwise be assessed separately.
While AI should never replace human judgment, it can give compliance teams better context to prioritise investigations and intervene earlier if the right foundations are in place.
Singapore’s latest operation demonstrates the value of information moving quickly between banks and law enforcement. The same principle should apply within financial institutions.
As scammers become more adept at making fraudulent situations resemble legitimate transactions, the ability to connect customer information, transaction activity and other risk signals will be paramount for banks trying to tell the difference.
The writer is head of Asia-Pacific solutions consulting at Fenergo