Gartner has predicted that by 2028, as many as one in four candidate profiles worldwide could be fake (HR Dive). The tools to make that happen are cheap and quick: a researcher at Palo Alto Networks with no image-editing experience built a fake job candidate convincing enough for a video interview in about 70 minutes (HR Dive).
This is no longer a curiosity. In January 2025 the FBI warned US employers that North Korean IT workers were stealing identities and using AI and face-swapping technology during video interviews to get remote jobs, then stealing data and extorting companies (National Law Review).
What a deepfake candidate is
A deepfake candidate is a real person interviewing behind a synthetic face, voice or both:
- Face swap. Software replaces the interviewee's face with another person's — often the person whose identity and CV they stole — in real time.
- Voice clone. The voice is changed to match the stolen identity, or to hide an accent.
- Synthetic identity. A face that belongs to nobody, matched to a made-up CV and fake references.
The goal varies: a salary for a role the person isn't qualified for, access to systems and data, or a foothold for fraud.
How common it is
Deepfake fraud attempts rose by more than 1,300% in 2024, according to Pindrop, and only 0.1% of people in an iProov study correctly identified every real and fake sample — while 57% believed they could (both compiled by HeroHunt). Gut feel is not a defence.
The warning signs
- Edges. Flicker or blur around the jaw, hairline, glasses or ears, especially when the head turns.
- Occlusion. When a hand passes in front of the face, the mask glitches, lags or disappears. Palo Alto Networks recommends asking the candidate to do exactly this (HR Dive).
- Lip-sync drift. The mouth and the audio fall slightly out of step.
- Lighting that doesn't move. The face stays evenly lit while the room's light changes.
- Limited head movement. The candidate avoids turning sideways or looking down.
- A virtual camera. The video comes from software (a "virtual webcam") rather than a physical camera.
- Mismatched details. Location, time zone, phone number or work history that doesn't add up, or the same contact details shared with other applicants.
What HR teams can do today
1. Make the candidate move. Ask them to turn their head to the side, pass a hand across their face, or hold their ID next to their face.
2. Check the camera source. Flag virtual cameras and unusual video devices before the round begins.
3. Add a second, independent camera. A face swap runs on one video feed. A phone camera filming the candidate from the side is a second feed that the swap does not control.
4. Capture ID and face at every round and compare them across rounds — the same checks that stop proxy interviews.
5. Verify identity documents and cross-check contact details against other applicants, as the FBI recommends.
6. Meet in person before the offer for sensitive roles, or at least run a short supervised final round. Gartner research reported by Computerworld found 72.4% of recruiting leaders now run at least one in-person interview to combat fraud (HeroHunt).
7. Limit access after hiring. Give new remote hires the least access they need until their identity is confirmed in person.
What not to do
Do not run automated face recognition on candidates without a clear legal basis and their informed consent — it creates privacy risk of its own. Use evidence for a person to judge, and keep it only as long as you need it.
How Oximus Pro helps
Oximus Pro flags virtual cameras and unusual devices in the device check, captures face and ID before every round, shows the panel the first-round reference photo beside the live video, and adds the candidate's phone as an independent side camera. People decide what the evidence means. Book a demo.
Frequently asked questions
How can you tell if a job candidate is a deepfake?
Ask them to turn their head and pass a hand in front of their face, watch for blurring at the edges and lip-sync drift, check whether the video comes from a virtual camera, and compare their face and ID across rounds.
Are deepfake job interviews common?
They are still a minority of interviews but rising quickly, especially for remote technical roles. The FBI has warned employers about organised groups using face-swapping in video interviews.
Can deepfake detection software stop fake candidates?
It helps, but detectors can be fooled and produce false alarms. Combining device checks, a second camera, identity checks at every round and an in-person step is more reliable.
What roles are most targeted by deepfake candidates?
Remote IT, software development and roles with access to sensitive data or financial systems are the most common targets.
