Candidate fraud detection tools turned into a real product category in about eighteen months. Greenhouse partnered with CLEAR for identity checks in June 2025. Ashby shipped fraud detection inside its ATS that September. Persona launched a candidate verification product in March 2026. If you hire remote engineers, someone has probably tried to sell you one of these.
Here's what each type of tool does, what it can't do, and how to decide whether you need one. I run a staffing company, so I'll also say plainly where a vendor like us fits and where it doesn't replace your own checks.
Key Takeaways
- Gartner surveyed 3,000 job candidates and found 6% admitted to interview fraud. It also predicts one in four candidate profiles worldwide will be fake by 2028.
- Candidate fraud detection tools split into three jobs: application-stage risk signals, identity verification (ID plus selfie), and interview-stage deepfake or proxy detection.
- Ashby describes its own system as "signals, not verdicts." That's the right way to treat every tool in this category.
- Identity verification proves the person is real. It doesn't prove they can do the job, or that they're the one who did the interview.
The three types of candidate fraud detection tools
1. Application-stage signals. Built into or bolted onto the ATS. Ashby's fraud detection launch flags things like location mismatches, identity verification issues and assessment anomalies, and Ashby's own line is "We provide signals, not verdicts." Gem's fraud detection agent, powered by Tofu, looks at email, phone, LinkedIn account age, IP and resume metadata, per Gem's announcement. Cheap, automatic, noisy.
2. Identity verification. The candidate takes a selfie and scans a government ID, and the tool matches them. CLEAR and Greenhouse announced this in June 2025, with the goal of confirming candidates are "not AI-generated fakes or impersonators." Persona launched Candidate Verification on March 11, 2026, with integrations for Ashby, Greenhouse and Workday. LinkedIn's verification uses CLEAR for government ID in the US and confirms a current workplace through a work email (LinkedIn Help). Nothing in it checks past employers.
3. Interview-stage detection. Tools that watch the video call for synthetic faces, voice changes and proxies. Pindrop, Reality Defender and Sherlock sell in this group. I compared them in interview cheating detection tools.
| Tool type | Stops | Doesn't stop | When it runs |
|---|---|---|---|
| Application signals | Bulk fake applications, reused contact details, location lies | A real person with a real identity and borrowed skills | On apply |
| ID + selfie verification | Fake identities, stolen documents (often) | Proxy interviewers, a real person who lies about experience | Before interviews or at offer |
| Interview detection | Deepfakes, voice swaps between rounds | A helper off camera, an assistant feeding answers | During calls |
| Your own process | Shallow skills, inconsistent stories | Nothing perfectly | Every stage |
What the threat actually looks like
Most fake candidates are individuals stretching the truth. The dangerous ones are organized. In June 2025 the Justice Department said one North Korean IT worker scheme alone reached more than 100 US companies, and agents searched 29 suspected laptop farms across 16 states. Christina Chapman, who ran one, was sentenced to 102 months for a scheme that defrauded 309 US companies using 68 stolen identities.
Note what those cases had in common: real stolen identities. ID verification helps against invented people. It's weaker against a real identity in the wrong hands. That's why the FBI's January 2025 guidance asks employers to verify identity while interviewing, at onboarding, and after.
How to choose
- Under 20 hires a year: skip the platform. Do ID verification at offer with any reputable vendor, and design interviews that require the candidate's own experience. Our candidate identity verification guide covers the cheap version.
- High-volume engineering hiring: turn on application signals in your ATS, add ID verification before the final round, and review flags by hand.
- Using a staffing vendor: ask which of these layers they run, at which stage, and get it in writing. The FBI's May 2024 alert tells employers to make sure third-party staffing firms run sound hiring practices, not to assume it.
At Ruzora, every engineer on our bench passes two separate evaluations: a graded coding challenge and an AI interview answered on camera. You see blind profiles first, and full identities come after the service agreement, which is the right moment to run your own ID check on anyone you're about to hire, including people we send.
A Concrete Version
A 90-person healthtech company hires 12 remote engineers a year and gets about 400 applicants per role.
- On apply: ATS fraud signals flag roughly the top risk slice for a recruiter to review. Most flags turn out to be VPN users and new LinkedIn accounts. A few are clear duplicates using the same phone number across names.
- Before the final round (about 6 candidates per role): ID plus selfie verification. One candidate drops out rather than complete it.
- Final round: a live walkthrough of the candidate's own past work, camera on.
- Day one: a short video call with the same person before laptop and repo access. The face on the call matches the ID from the final round.
The tools filtered the noise. The decisions came from a person checking the same human at four points.
The Honest Counterpoint
These tools add friction, and friction costs you candidates. Senior engineers with three offers won't always scan a passport to take your first screen. ID checks before any interview also feel like suspicion to honest people. Put the heavier checks later, where the candidate has a reason to stay.
False positives fall unevenly too. VPNs, shared apartments, recently created accounts and travel all look like risk signals and all have innocent explanations, more often for international candidates. If a tool's flag ends a process automatically, you'll lose good people and never know.
Frequently Asked Questions
What are candidate fraud detection tools?
Software that flags fake or misrepresented candidates. The main types are application-stage risk signals in your ATS, identity verification that matches a selfie to a government ID, and interview-stage detection of deepfakes, voice changes and proxies.
Do candidate fraud detection tools stop North Korean IT workers?
They help, especially ID verification and checks on location and payment details. But those schemes often use real stolen identities, so you also need consistency checks across interviews, onboarding and the first weeks of work. See North Korean IT workers screening.
Should a staffing agency run candidate fraud detection for us?
It should run its own checks and tell you which ones, at which stage. You should still verify identity yourself before granting access. A vendor's process is one layer, not a replacement for yours.
The Bottom Line
Candidate fraud detection tools are worth it at volume, as long as you treat their output as signals and keep a person on every decision. Check the same human at several points, and verify identity before access, not after. To start from engineers who already passed two separate evaluations, see our bench or read how we vet.
Roberto Espinoza is CEO of Ruzora, which helps US startups hire pre-vetted senior LATAM engineers, with a vetted shortlist in 72 hours. See available engineers.
