Fraud Risk Classification

While Identity Verification asks "is this the right person?", Fraud Risk Classification addresses the probabilistic variable: what is the real risk of this transaction turning into a financial loss for your business?

What it does

The engine cross-references the face and the government identifier with additional metadata to identify data inconsistencies, history of third-party document use, behavioral anomalies, and multi-account activity.

Adoption advantage: can be consumed through our SDKs or via API only.

Inputs

A process created via Create Process (Web & Native) or Create Process (API).

Required:

Person's selfie — in API integrations you send it as imageBase64; in Web integrations the capture is performed by Unico.
Primary identifier (subject.code in API) — government ID, email, phone number, etc.

Optional (Highly Recommended):

clientReference (subject.clientReference in API) — becomes required when Multi Accounts is enabled.
Extra metadata: email, phone, demographic data (name, date of birth, gender, etc.) — these are the essential inputs for a complete behavioral model analysis.

Possible responses

Response	Meaning
`YES`	Signs of fraud associated with this face.
`INCONCLUSIVE`	No signs of fraud associated with this face.

This table represents the individual response of the capability.

API enum values

In the Web & Native contract response, these map to TRUST_RESULT_YES (YES) and TRUST_RESULT_INCONCLUSIVE (INCONCLUSIVE) in the authenticationInfo.identityFraudstersResult field. See Get Process for the full enum table.

There is no NO response

Fraud Risk Classification does not return NO — the absence of evidence is represented as INCONCLUSIVE, not as a confirmation that there is no risk.

Interpreting the result

The behavioral assessment returns one of 5 possible values. When status = 3 (Completed), check riskLevel.result:

Value	Meaning / What the engine found	Recommended action
`approved`	No evidence of fraud associated with this face.	Proceed with the standard approval flow.
`reproved`	Maximum evidence of fraud and risk. There is a very high probability that this transaction will result in financial loss, due to strong association with identity theft or recurring use of third-party identities.	Automatic block or rejection.
`risk-critical`	High fraud risk. Indicates critical inconsistencies, such as a severe mismatch between registration data (age/gender) and facial biometrics, pointing to a high risk associated with fraud.	Automatic rejection or strict credit policy adjustment.
`risk-high`	Potential fraud. This signal frequently reflects unusual recent activity or network hyperactivity.	Manual review by the analysis desk (back-office).
`inconclusive`	There is not enough information mapped in our network to determine that the transaction carries risk.	Proceed with the standard approval flow.

Multi Accounts

Answers whether the captured face is already linked to another clientReference within the same clientReferenceSegment. See the Multi Accounts page for the possible responses, requirements, and all the capability's information.

Availability

Surface	Supported
SDK (Android, iOS, Flutter)	✅
Web (iFrame, Redirect)	✅
API (headless, no SDK)	✅

Valid combinations

Fraud Risk Classification appears in flows that include the trust suffix:

idlivetrust, idtrust, idchecktrust, idchecktrustdocs, idchecktrustsign, idchecktrustdocssign, idtrustdocs, idtrustsign, idtrustdocssign, idtokentrust.

For the full matrix, see Available flows.

What it does​

Inputs​

Possible responses​

Interpreting the result​

Multi Accounts​

Availability​

Valid combinations​

Products that use this capability​