Use case 03
Authorised payment scams
The genuine customer is acting. The payment is still the fraud.
- AMOUNT_OUTLIER
- FIRST_TIME_RECIPIENT
- HIGH_VALUE_LOW_ASSURANCE_CHANNEL
- MULE_DORMANT_TO_EXPLOSIVE
- MULE_UNRELATED_SENDERS
- RECIPIENT_FIRST_TIME_SENDER_CONCENTRATION
- RECIPIENT_MANY_SENDERS_1H
- RECIPIENT_NEW_ACCOUNT
The problem
In an authorised scam the identity checks all pass, because the real customer is the one pressing send. Systems that fold identity assurance into a single score are pulled towards allow exactly when they should not be.
How Cenvoriq approaches it
Identity Trust stays high because the genuine customer is acting; Transaction Risk, Recipient Risk, Mule Risk and Scam Risk describe what is happening and to whom, and are never summed with it. The recommendation ladder acts on the strongest dimension.
Status in the current release
Built and tested; shown in demo scenario F, where Identity Trust is 100 and the recommendation is still manual review.
- Amount far above the sender's own history
- First payment to this recipient
- High value over a low-assurance channel
- Recipient collecting from many first-time senders
- Recent account recovery on the sender
Deployment
Measured in SHADOW before it touches a payment
A design partner replays its own tokenised history first, then mirrors a live stream. The engine scores this pattern and records the recommendation; the applied action is always allow and the institution’s own systems keep every control.