Trust every application. Verify every hire.
Integrity signals.
Stage · ApplyGenerative AI made fabrication and synthetic identity cheap and fast.
candidate profiles projected to be fake by 2028
year-over-year rise in DPRK operatives hired at Western firms
of fraudulent hires receive active corporate credentials before they are detected
of candidates admit to interview impersonation
Where candidate fraud can surface.
Fabricated information
- Invented or inflated employers, schools, titles and dates
- Contradictory employment histories across applications
- Suspicious document metadata such as PDF author and creation tool
- Bot-generated or AI-fabricated resume content
Impersonation
- Someone else sits the phone screen or the interview
- Synthetic or stolen identities used to apply
- State-sponsored infiltration, including the DPRK IT-worker network
- A different person on day one than the person you interviewed
| TOOL | LIMIT | WHY IT FAILS |
|---|---|---|
| Background checks | Post-offer, one-time | Historical only, runs after weeks of investment, no recruiter integration |
| Point verification tools | Application stage only | One checkpoint, no continuity across stages |
| Manual review | Inconsistent and reactive | Does not scale to application volume |
Each of these verifies a moment. Fraud lives in the handoffs between them. A fabricated skill set clears the resume review. A different person sits the technical interview. A third address receives the laptop. Each tool did its job. Nothing was watching the chain.
Integrity signals. Each stage. One record.
Each stage adds new information to the same record, so the application integrity signals evolve as the hiring process progresses.
01
apply
Resume and document credential and metadata signals. Contact data signals across IP, phone and email.
02
AI Interview 360
Profile consistency between the live interview and the submitted application. Identity verification, configurable by customer. Interaction indicators from proctored interviews. Skills validation against a customer-defined rubric.
03
Human interview
Consistency check with cross-stage signal continuity.
Detecting a polished answer is not the same as verifying a skill.
Reading speech patterns can flag a borrowed answer. It cannot verify a skill. Vocabulary, filler words and the pause before a reply show how someone answers, not necessarily what they can do. Eightfold’s 360 Interview tests the skill itself. A short structured interview, scored against a rubric your team defines, in coding, case and scenario formats. The candidate’s profile then shows which skills were validated and which were only claimed.
- Rubric defined by you, not inferred from language patterns
- Claimed skills reconciled against the submitted application, round by round
- Validation status written to the profile with evidence and carried forward
What we check, named in plain language.
Integrity signals fire the moment a candidate applies, then shift from static to interaction indicators as the process moves forward.
| STAGE | SIGNAL | WHAT IT CATCHES |
|---|---|---|
| Apply | Resume | Document metadata including PDF author, creation tool and template reuse. |
| Apply | IP address | Session Geography Signal. Anonymized submission origin indicators. |
| Apply | Phone number | Duplicate profiles sharing a number. VoIP, toll-free and non-personal line types |
| Apply | Email address | Disposable and temporary domains. |
| AI interview | Profile consistency | Compares interview responses on titles, employers, dates and claimed skills directly against the submitted application. Logs the specific field and discrepancy, not a score |
| AI interview | Interaction Indicators | Timestamped signals added to the running record, for human review with full context |
Signals inform. Humans decide.
No automatic rejection.
Recruiters decide. No application is auto-rejected.
Explainable signals.
Named and graded.
Optional ID verification.
Opt-in, consent-based, not mandatory.
Auditable and explainable.
Transparent logic, monitored for disparate impact.
Enterprise security, built in from day one. Not bolted on after the fact.
Hiring risk doesn’t stop at recruiting.
Talent Acquisition
Stop losing weeks to candidates who were not what their application suggested. Explainable signals instead of gut feel, and a faster path to genuine finalists.
Security and IT
Compliance and Legal
Hire with intelligence. Hire with confidence.
No signal is lost.
Each integrity signal is logged to a running application record. Nothing falls through the cracks between stages.
Anomalies surface early.
Flagged at the earliest possible stage, before pipeline time is wasted.
Explainable, not opaque.
Each signal is named, graded and described.
Humans own each hire.
No application is auto-rejected. Signals inform. Recruiters decide.
Candidate fraud, answered.
Any deliberate misrepresentation in the hiring process. It falls into two categories. Fabrication is misrepresenting credentials, work history or skills. Impersonation is one person presenting as another, in an interview or on day one. Gartner projects that one in four candidate profiles will be fake by 2028, and 6% of candidates admit to interview impersonation.
We address impersonation risk by incorporating identity verification into the hiring process. Through integrations with ID.me and CLEAR, teams can verify that the person behind an application is who they claim to be. This provides an additional layer of protection against synthetic identities and other forms of impersonation, including deepfakes.
Through validation rather than inference. The 360 Interview tests claimed skills against demonstrated ability, scored to a rubric your team defines, and writes the result as part of the hiring record. Language analysis can suggest an answer was not the candidate’s own. Only validation tells you whether they can demonstrate the skills required for the role.
No signal rejects anyone. Eightfold surfaces anomalies with explanations and leaves the decision with your team. Recruiters can dismiss a flag with a note or add their own. Probabilistic signals surface only at high confidence, and human review is a design principle rather than a fallback.
Yes. Any new application triggers computation regardless of where it originated.
Yes. Eightfold consolidates signals from your existing providers into one integrity record through the app platform. We integrate natively with ID.me and CLEAR.
Out-of-the-box rules are active on enablement, including a first pass across your existing active applications. Custom rules take as long as you want to spend writing and testing them.