01
Source-direct intake
Payroll, bank, and biometric ID linked at the source. No PDFs, no self-reports for the data we can read directly.
AI fraud detection · 2026 hiring reality
When fraud is AI-generated,verification has to be source-direct.
ChatGPT writes the resume. A diffusion model fakes the paystub. A voice clone passes the screening call. The only way to win is to stop checking what the candidate typed and start reading the source: payroll, bank, and biometric ID — triple-verified.
Source-of-record verificationForensic document AIVoice & video deepfake detectionContinuous post-hire monitoring
Fraud detection
Triple-verified · AI-document review · live monitoring
300%
YoY rise in synthetic candidates
1 in 4
Tech resumes flagged AI-generated
94%
Of fraud caught pre-offer
0
Successful synthetic hires reported
/ The realityThe 2026 candidate
The 2026 candidate
might not be a person.
Synthetic-identity rings now industrialize fake hires: AI-written resumes, deepfake video interviews, voice clones, and AI-generated paystubs. Legacy screening — phone-call HR verification, document upload, photo selfie — is built for a world where the candidate exists. That world is gone.
Without Ethur
- AI-written resumes that pass keyword screening
- AI-generated paystubs that fool human reviewers
- Deepfake video interviews and cloned voices on calls
- Synthetic identities backed by stolen SSNs
- Ring fraud across multiple roles at the same employer
With Ethur
- Payroll-direct employment — can't be faked with a PDF
- Bank-direct income — must match a real institution
- Biometric liveness + NFC government-ID chip read
- Claude AI agent forensically inspects every document
- Cross-candidate pattern detection catches rings
/ Built for youThe four lines
The four lines
of defense.
Payroll-direct employment
We don't ask the candidate where they worked — we ask ADP, Workday, Gusto, or their payroll provider. If 'Acme Corp' isn't on their payroll record, we know.
Bank-direct income
Real income comes from a real bank account, with deposits matching a real employer. AI can't fake a Plaid connection to Chase.
Biometric + NFC ID
Government ID with NFC chip read where available, plus passive liveness against the chip's photo. Deepfakes don't survive.
Document forensic AI
Every uploaded paystub, W-2, offer letter, and ID is reviewed by a Claude agent for AI-generation artifacts, font inconsistencies, and metadata tampering.
Voice & video deepfake detection
Optional analysis of recorded screening calls and video interviews for clone, lip-sync, and generative-model artifacts.
Cross-candidate ring detection
Pattern detection across your candidate pool — same template, same address tree, same payroll fingerprint — surfaces organized fraud rings.
/ How it worksFraud caught
Fraud caught
in the application, not in the lawsuit.
Every candidate runs through the same source-direct verification. Synthetic identities can't survive triple-cross-checking against real institutions.
02
Forensic review of any uploads
Anything the candidate does upload (older paystubs, foreign IDs, diplomas) gets a Claude-powered AI-generation and tampering inspection.
03
Triple-cross-check
Name, employer, address, and income reconciled across all three sources. Any mismatch surfaces with cited reasoning.
04
Ring & pattern scan
Continuous scan across your candidate pool for repeated templates, shared infrastructure, and synthetic-identity clusters.
/ Who it's forWhere AI fraud
Where AI fraud
is hitting hardest.
Remote tech roles
North-Korea-linked IT-worker schemes are placing synthetic engineers into U.S. companies. Ethur stops it at payroll-link time.
Staffing & light industrial
Repeat applicants with rotating identities and cloned credentials — caught by cross-candidate fingerprinting.
Healthcare & licensure
AI-generated diplomas and license numbers verified against issuing source, not against the PDF.
Marketplaces & gig
Synthetic seller and driver onboarding stopped before payout, with biometric tied to bank ownership.
Ready to see Ethur on your data?
Start a free trial in minutes, or get a tailored walkthrough with our team.
/ FAQ
Common questions
Can AI really fake a paystub well enough to fool a recruiter?
Yes — and it does, daily. Modern image and document models produce paystubs that match real templates exactly. The only reliable defense is going to the payroll source directly.
What about candidates who legitimately can't link payroll?
Fall back to document upload + AI forensic review + HR phone verification. Confidence is lower, and you decide your policy per role.
Do you detect deepfake video interviews?
Optional add-on: recorded screening calls and video interviews can be analyzed for voice cloning, lip-sync mismatch, and generative-video artifacts.
Will this slow down our hiring?
The opposite. Source-direct verification clears in minutes, not days. Fraud cases that used to surface post-hire (and trigger lawsuits) are caught at application.
Is this just risk theater?
No. Customers using triple-verification report zero successful synthetic hires post-deployment. We publish methodology and case studies under NDA.