Deeptrack
Request an assessment
Blog

Blog

The Candidate Isn't Real: Deepfakes in Hiring

Gartner projects one in four job candidates will be fake by 2028. Here is what the data says about deepfake candidates, how they get hired, and what African employers can do now.

Brian Koyundi · · 7 min read

A security company hired a software engineer who passed four video interviews and a background check. Within 25 minutes of receiving his work laptop, he began loading malware. The engineer was not who he said he was (KnowBe4 case, via Cloud Security Alliance).

That case is no longer rare. Gartner projects that by 2028, one in four job candidates worldwide will be fake (CNBC). Real-time face-swaps and cloned voices now let a stand-in pass the one step most employers still trust without checking: the live video interview.

The numbers: fake candidates are already in the pipeline

Across surveys, security firms and enforcement cases, the pattern is consistent: candidate fraud is common, much of it is caught only after hiring, and deepfakes are now part of the toolkit.

HR professionals who have personally experienced candidate fraud98%Hypr, 2026 State of HR Identity Fraud Detection, via Biometric Update
Candidate-fraud cases discovered only after the person was hired42%Hypr, via Biometric Update
Companies that took 1–3 months to resolve a case1 in 4Hypr, via Biometric Update
Applicants in one remote hiring pipeline showing clear signs of fraud1 in 6Pindrop, own pipeline
State-linked applicants who used real-time deepfakes in live interviews1 in 4Pindrop, own pipeline
Engineering candidates where a different person attended the technical interview6%Pindrop, one software engineering role
Confirmed remote-worker fraud incidents investigated in a year, up 220%320CrowdStrike, year to June 2025, via CSA research note
Interviews run by 130+ fake identities, at 5,000+ companies6,500+Okta, via CSA research note
Suspected state-linked applicants blocked by one employer since April 20241,800+Amazon, via CSA research note
Job candidates worldwide expected to be fake by 20281 in 4Gartner, via CNBC

A note on the figures: Hypr and Pindrop sell fraud-prevention tools and report their own data, and "candidate fraud" in surveys includes fake CVs as well as deepfakes. The direction is clear even where the exact size is not.

How a deepfake candidate gets hired

The identity on paper is usually real, often stolen, so document and background checks pass. The fake appears on camera.

  1. Apply. An AI-tailored CV, a stolen or synthetic identity and an AI-generated photo get past the applicant tracking system.
  2. Screening call. A cloned or accent-shifted voice handles the recruiter's phone screen.
  3. Video interview. A real-time face-swap, or a proxy who sits the technical test, passes the panel. In one Pindrop pipeline, 79% of candidates for an MLOps role did their technical interview with the camera off.
  4. Background check. The stolen identity is genuine, so education, employment and criminal-record checks come back clean.
  5. Onboarding. A different person starts work, or the laptop is shipped to a "laptop farm" where it is operated remotely. One US facilitator received hardware from 300+ companies.
  6. On the job. Some operators hold several jobs at once. Since late 2024, some dismissed workers have copied source code and demanded payment to keep it private.

Most employers check steps 1 and 4 thoroughly. Steps 2, 3 and 5, the live calls where deepfakes operate, are rarely checked at all.

Why African employers should pay attention

Most reported cases come from US tech firms, but the conditions that enable them are growing fastest in Africa: remote hiring at scale, and identity checks that spoofing already beats.

  • Remote work is Kenya's growth engine. Kenya's digital economy supports 350,000+ jobs, more than 40,000 of them in BPO and global business services, and the government is pushing the sector into AI operations, software and cybersecurity (TechAfrica News, Sep 2026). Each of those roles is hired, and often worked, over video.
  • Foreign clients will set the bar. BPOs and talent platforms that place African staff with US and European companies will be asked to prove the person on the call is the person on the contract.
  • Spoofing already beats identity checks here. Smile ID's 2026 Digital Identity Fraud Report, built on more than 200 million checks in 35 countries, found about 65% of digital fraud attempts in West Africa involved spoofing or failed face-matching, and it detected over 100,000 injection-style attacks a month in 2025 (BusinessDay, Mar 2026). The tools that beat bank onboarding work just as well on a hiring selfie.
  • The law treats faces and voices as sensitive data. Kenya's ODPC Guidance Note on Biometric Data requires explicit consent, a DPIA filed 60 days before processing, and a copy of the data stored in Kenya. Any check employers add must be built for that.

What employers can do now

The fix is to verify the person, not just the paperwork, at every step where they appear live.

  1. Keep cameras on. Make video mandatory for every interview round, and treat repeated camera or connection problems as a signal, not a nuisance.
  2. Check liveness, not just a selfie. A liveness and deepfake check on a short video at application catches the easy fakes before a recruiter spends an hour on them.
  3. Scan the live interview. Deepfake detection inside Zoom, Google Meet or Teams flags face-swaps and cloned voices while the call is happening.
  4. Make sure the same person shows up. Re-verify identity on the first day and on the onboarding call, and match it to the person interviewed. This closes the HR-to-IT handoff gap.
  5. Ship hardware carefully. Confirm on camera that the new hire received the laptop, and flag any change of delivery address after the offer.
  6. Grant access in stages. Separate approvals for code, production and admin rights, with monitoring from day one.
  7. Get consent right. Under Kenya's Data Protection Act and GDPR, faces and voices are sensitive data: get explicit consent, offer a non-biometric alternative, and keep the data only as long as you need it.

How Deeptrack helps

Deeptrack builds deepfake detection for faces and voices, used today by financial institutions to stop fraud at onboarding. The same engine can check candidates where deepfakes operate: the recorded video at application, the live interview, and the onboarding call. Hiring checks are available for pilots with employers, BPOs and screening partners. Request an assessment to run one on your next hiring round.

Sources