Scam complaints are Singapore's money laundering problem

What the evaluation found

The Financial Action Task Force and the Asia/Pacific Group assessed Singapore following an onsite visit in July 2025, and published their evaluation in May 2026.

Singapore came out of it well. It moved from enhanced to regular follow-up, its best result to date, and the report describes a competent and coordinated regime.

The finding that should interest a bank is narrower. Fraud, particularly scams and other cyber-enabled financial crime, much of it run by syndicates based overseas, is named among the country's prominent money laundering challenges.

Why that changes how a scam claim looks

In most banks scam handling and anti-money laundering sit in different teams with different systems and different reporting lines.

The scam desk is measured on customer outcomes and turnaround. The financial crime team is measured on alerts, reviews and reports filed. They rarely share a case.

But if fraud is a principal source of laundered money, then a scam complaint is often the first visible edge of a laundering network. The customer telling you they were tricked is describing the entry point.

Treating that complaint purely as a service matter means the intelligence in it goes nowhere.

The newer police powers point at the same networks. The Scams (Countermeasures) and Other Matters Bill adds orders against accounts used to enable scams, including accounts run by money mules.

The scale is large even in a good year. Scam cases in Singapore fell 27.6% in 2025 to 37,308, and losses fell 17.9% to about S$913.1 million, according to the Singapore Police Force. In the same year more than 7,000 money mules and scammers were investigated and more than 940 were charged in court.

11,189 and 682

The report counted 11,189 money laundering investigations, with 682 resulting in prosecution. That is under one in ten.

Read that carefully, because it is easy to misread. It is not evidence that the investigations were wrong, and Singapore's overall assessment was strong.

It does say something about what it takes to turn a case into an outcome. Evidence has to be complete, traceable and assembled well enough to survive scrutiny long after the event.

That standard is set well before anything reaches a prosecutor. It is set on the day the bank works the case.

What a bank can actually change

A bank does not control prosecution rates. It controls the quality of what it hands over, and how quickly.

• Whether the fund flow out of a victim's account was traced while the trail was fresh
• Whether the receiving accounts were examined as a network rather than one at a time
• Whether the scam complaint and the monitoring alerts on the same accounts were ever connected
• Whether the evidence pack is complete, or complete apart from the parts nobody had time for
• Whether all of it can be reconstructed a year later by someone who was not there

None of that is exotic. It is careful, repetitive work, done under time pressure, at volume, which is exactly the kind of work that degrades first when the queue grows.

Consistency is a volume problem

Execution quality is not mainly a function of how good your analysts are. It is a function of how many cases each of them has to get through today.

Give a capable analyst twenty cases and the work is thorough. Give the same analyst eighty and something gets shortened, and it will be the part that takes longest and shows least - which is usually the evidence gathering.

Results reported across one active deployment show the scale of that cost: case response ran at about three days before evidence assembly stopped being manual, and almost none of that was decision-making.

The connection worth building

The practical version of the finding is this: a scam complaint and a transaction monitoring alert on the same account are two views of one event.

Most operating models keep them apart. Different queue, different team, different system, and no reliable way to notice that both fired on the same customer in the same week.

CaseLEA works the preparation layer under both. It gathers the account history, the counterparty picture and the transaction trail once, and presents them to whichever analyst is looking.

The analyst still decides. But they decide with the whole picture rather than their half of it, and that is a change worth making regardless of what any evaluation says.

What to measure on this specifically

Source: FATF and APG mutual evaluation of Singapore, published May 2026.

For your own operation the useful counterpart is how often a scam complaint and an alert on the same customer were worked by the same person, or at least with sight of each other. Most banks cannot answer that, which is the finding.

What the evaluation does not say

The mutual evaluation is worth reading precisely, because it gets quoted loosely.

The evaluation does not say that Singapore's regime is failing. It moved the country from enhanced to regular follow-up, its best result to date, and describes a competent and coordinated system.

It does not attribute the gap between 11,189 investigations and 682 prosecutions to any particular cause, and reading a failure into that ratio would be inventing a finding.

What it does establish is that fraud and scams sit among the prominent money laundering threats to the country. Everything else in this post is our analysis of what that means operationally, not the report's.

The first thing to change, if you change one thing

Make it possible to see, on one screen, that a scam complaint and a monitoring alert exist on the same customer.

That sounds trivial and it is usually blocked by system boundaries rather than by policy. The scam desk works in one tool, financial crime in another, and no reliable join exists between them.

Resolving it does not require either team to change how they decide. It requires the evidence to be gathered once and made visible to both, and it is the change with the shortest path between effort and effect in this whole area.