FraudShield Launch: 15 More Signals for Improved Risk Scores

Fraudsters adapt to the checks they know recruiting teams use. This FraudShield release expands the evidence behind every risk score so teams can investigate suspicious applicants without forcing more work or friction onto every genuine candidate.
15 Additional Fraud Signals
The machine-learning system underneath FraudShield now evaluates 15 additional signals across the Endorsed platform and Chrome extension. The release expands:
- Identity Verification: New data sources provide more detailed evidence about whether an applicant's internet identity is consistent with the person they claim to be.
- Confidential Signals: Twelve new indicators surface patterns correlated with fraud without publishing a checklist that malicious applicants can tune around.
- Phone Ownership and Validity: Recruiters can see whether the listed owner appears related to the applicant and whether the number is valid.
- Risk Interpretation: The model has more context for weighing each signal as part of the overall result.
Why Some Signals Stay Confidential
Earlier identity checks—such as a new LinkedIn account or a document match—could be useful on their own. But fraud rings study common controls, then change their behavior or place a real person in front of a stolen or fabricated work identity.
The confidential category makes that adaptation harder. Recruiters still receive an understandable risk level and enough context to conduct a fair review, while the underlying detection methods remain protected.
Evidence for Review, Not an Automatic Verdict
A single signal is not proof that an applicant is fraudulent. VoIP numbers, sparse online histories, and identity mismatches can all have legitimate explanations. FraudShield combines the full pattern and lets reviewers mark a result as fraud, not fraud, or unsure.
That approach helps teams focus their investigation on the highest-risk applications while continuing to review genuine candidates quickly and consistently.
The expanded signal set is live for Endorsed customers. Supported Chrome extensions receive the model update automatically.