93.7%
Recall — ML-based AI agent catching laundering cases
$20B
Estimated money laundering losses in Malaysia, 2024
~50%
Potential reduction in compliance costs with AI-driven AML
Introduction
The financial crime landscape is not standing still. Laundering techniques are becoming more layered, more automated, and more sophisticated. And yet, for most financial institutions, the core detection process has barely changed: flag suspicious transactions, investigate, close. Repeat.
The problem with that loop isn’t effort, it’s architecture. Today’s AML systems are built to answer one question: did we catch the bad ones? But that framing leaves an equally important question on the table: how many good ones did we wrongly flag?
Both errors carry real costs. A missed laundering case represents a compliance failure with regulatory and reputational consequences. An excessive false alarm rate drives up operational costs, burns out compliance teams, and creates investigation backlogs that ironically slow down the detection of genuine cases.
“The goal isn’t just to catch more money laundering. It’s to catch more, while investigating less.”
Where single-agent systems fall short
In recent research published by Cybiant, four AML approaches were benchmarked against each other: a traditional rule-based model, a CatBoost machine learning model, and two AI agents augmenting each of those benchmarks respectively.
The results were clear: integrating an AI agent as a second layer meaningfully improved detection. The ML-based AI agent achieved a Recall of 93.7% — a 7.8 percentage point improvement over the standalone ML model. But optimising for Recall comes at a cost: the False Alarm Rate also rose by 18.1 percentage points. More signal, but more noise alongside it.
A single agent doing both jobs — catching everything suspicious while filtering out what isn’t — is inherently difficult. It’s asking one system to be simultaneously aggressive and conservative.
The double-agent concept: divide and conquer
Double-agent AML architecture
Transactions
Raw data stream
ML Model
CatBoost baseline
Primary Agent
Catch false negatives
· Max recall
Secondary Agent
Filter false positives
· Max precision
Clean Alerts
For investigators
Primary: Aggressive, high-recall — surfaces every suspicious transaction the model missed
Secondary: Conservative filter — reviews every alert before it reaches an investigator
The solution lies in separating those responsibilities. Rather than asking one agent to balance two competing objectives, a Double-Agent Framework assigns a dedicated agent to each task.
The Primary Agent is configured to be aggressive and high-recall — catching every suspicious transaction the benchmark model missed. The Secondary Agent then takes on the filtering role, reviewing every alert and evaluating whether each one genuinely warrants investigation.
“Two agents, two jobs. One to catch. One to validate. Together, they close the gap that neither could close alone.”
This architecture mirrors how high-performing human investigation teams actually operate. A first analyst surfaces potential cases; a senior reviewer applies judgement before escalation. The difference is scale — an AI-powered double-agent system can operate at the speed and volume of digital transaction flows that no human team can match alone.
Why this matters now
According to figures cited in Cybiant’s research, money laundering in Malaysia alone accounted for an estimated 5.04% of GDP in 2024 — over USD 20 billion in losses. At the same time, the annual cost of compliance has reached nearly USD 2 billion. AI-driven AML approaches have the potential to reduce those compliance costs by close to 50%.
The dual cost of AML failure
But cost efficiency without detection accuracy is a false economy. The Double-Agent Framework is designed to deliver both: lower operational burden and higher detection performance, simultaneously. And when the false positive rate is reduced upstream by the Secondary Agent, the alerts that reach investigators are more likely to be real — changing the nature of the work from managing volume to applying expertise.
What comes next
The Double-Agent Framework is the natural next step in Cybiant’s AML research programme. The foundations are already proven — the Primary Agent architecture works and the performance benchmarks are established. The next phase involves developing and testing the Secondary Agent: evaluating how effectively it reduces false positive rates without sacrificing the high recall the Primary Agent delivers.
AML is not a problem that gets solved and stays solved. The Double-Agent Framework isn’t just a research concept — it’s a practical architecture that financial institutions can begin designing for today.
“One agent catches what the model misses. The other ensures your team investigates what actually matters.”
Read the Full Study
The complete paper covers our full methodology, statistical analysis, model architectures, and extended findings — including a detailed breakdown of False Positive trade-offs across all four approaches.
Download the Research Paper
Free access. We may occasionally share relevant research from Cybiant — unsubscribe any time.