Detecting Complete MULE and AML Rings in a Single Rule — Ayzal AI FRM
Money laundering is a sequence, not a single event. Learn how to chain five specific composite steps in Ayzal AI FRM to detect a complete mule lifecycle: receiving funds, fanning out to multiple accounts, looping back, and draining the balance.
Detecting Complete MULE and AML Rings in a Single Rule
Traditional fraud rules look for a single bad transaction. But a mule account does not look bad on a single transaction. It looks normal. It receives money, and then it sends money.
To catch money laundering and mule networks, you have to look at the sequence of events. You need to detect the inflow, the fan-out to multiple beneficiaries, the loop back, and the total volume drained.
Here is how to build a complete MULE and AML Ring detection rule using the Ayzal AI FRM composition engine.
The Anatomy of a Mule
A mule account is a funnel. The attacker uses it to move funds from a compromised source to a safe destination. The lifecycle usually follows this pattern:
- The Inflow: A large sum is credited to the account.
- The Fan-Out: The funds are immediately split and sent to multiple different accounts (beneficiaries).
- The Loop Back: Some of the beneficiaries send a portion of the funds back to the original account, creating a circular flow to hide the trail.
- The Drain: The account continues to move large volumes of money, acting as a hub.
Building the Composite Rule
To detect this entire lifecycle in a single alert, we chain five specific steps together using an AND logic.
Rule Type: composition
Step 1: The Inflow
- Step Type:
field_compare - Field:
nature - Operator:
eq - Value:
credit
Step 2: The Outflow
- Step Type:
field_compare - Field:
nature - Operator:
eq - Value:
debit
Step 3: Fan-Out to Multiple Accounts
- Step Type:
count_distinct - Source Entity:
source_account - Target Entity:
dest_account - Threshold:
>= 3 - Time Window:
86400(24 hours)
Step 4: The Loop Back (AML Ring)
- Step Type:
entity_chain - Path:
source_account->dest_account->source_account - Threshold Endpoints:
2
Step 5: High Velocity Drain
- Step Type:
sum_amount - Entity:
source_account - Threshold:
>= 100000 - Time Window:
604800(7 days)
How the Backend Evaluates This Rule
When a transaction arrives, the engine does not just look at that one row. It traverses the entity graph and historical data:
- It checks if the transaction is a credit (Step 1).
- It looks back and sees the account has made multiple debit transactions (Step 2).
- It builds a graph of the account's beneficiaries and sees it sent money to 3 or more distinct accounts (Step 3).
- It traverses the graph and sees that one of those beneficiaries eventually sent money back to the original account (Step 4).
- It checks the total volume and sees the account moved over 100,000 in 7 days (Step 5).
If all five conditions are met, it generates a single MULE / AML Ring Alert.
Example Walkthrough
Imagine a mule account ACC-1001:
- It receives 50,000 from a compromised account. (Step 1 passes)
- It sends 10,000 to
ACC-2001, 15,000 toACC-2002, and 20,000 toACC-2003. (Steps 2 and 3 pass) ACC-2001sends 8,000 back toACC-1001. (Step 4 passes)- The total money moved by
ACC-1001in 7 days is 150,000. (Step 5 passes)
The system fires a single alert for ACC-1001, showing the analyst that this account is a mule acting as a ring hub.
See It in Your Own Data
Ayzal AI FRM combines real-time wallet transaction monitoring, entity graph detection, and per-product ML models to stop mule networks before they cash out.
