Deepfake Interview Detection Guide for Hiring Teams
By Keara Garrett · Updated November 4, 2026
Deepfake interview detection is the set of checks that confirm the person on your video call is the person behind the candidate profile. No single tool does this reliably. What works is a layered process: verify identity before the interview, watch for signs of manipulation during it, add one live step that is hard to fake, and confirm identity again before you grant access.
This guide is for HR and talent acquisition teams, procurement leads and tech leaders who hire remote talent in Canada and the US. It covers the warning signs, the checks you can run on a call, a step-by-step verification process and the limits of detection software.
Key takeaways
- A deepfake interview is one form of candidate fraud. Others include stolen identities, stand-in interviewees and invented work history.
- Gartner forecasts that by 2028, one in four candidate profiles worldwide will be fake.
- Detection software is useful but imperfect. Process controls do more of the work.
- Remote roles with access to systems, code or customer data carry the most risk.
- Tell candidates what you check and why. Transparency protects trust and keeps you on the right side of privacy rules.
What is a deepfake interview?
A deepfake interview is a video interview where AI alters or replaces a candidate's face, voice or both. The tools are cheap. Penn Career Services describes a demonstration in which a cybersecurity advisor built a fake candidate with about $50 of software that swapped his face and voice in real time.
Not every fake candidate uses AI. Some send a stand-in to answer questions. Others use a real person's stolen identity and documents. The checks in this guide work against all three.
How common is deepfake interview fraud?
Nobody has a precise number. The best public data points are these:
- In a Gartner survey of 3,000 job candidates, 6% admitted to interview fraud. That means posing as someone else, or having someone else pose as them. Self-reported numbers can undercount.
- Gartner predicts that by 2028, one in four candidate profiles worldwide will be fake. This is a forecast about candidate profiles, not about the people on your calls.
- In a 2024 Gartner survey, only 7% of recruiting leaders were using identity document validation software.
- In June 2022, the FBI's Internet Crime Complaint Center warned of rising complaints about deepfakes and stolen personal information used to apply for remote jobs. Some of those roles had access to customer data, financial data and corporate IT systems.
Which roles are most at risk?
The FBI's warning named remote IT, programming, database and software roles. The common thread is access. If a role gives someone entry to code, customer records or internal systems on day one, it deserves the strongest checks.
Fast onboarding adds risk. That applies to contract and consulting engagements as much as to employees, because there is less time for checks before work starts.
What are the signs of a deepfake in a video interview?
No single sign proves anything. Real people have poor webcams, weak connections and bad lighting. Treat these as prompts for another check, not as grounds to reject a candidate.
Video signs
- Lip movements that drift out of sync with the audio
- Blurring or flickering around the hairline, ears, glasses or jaw
- Lighting on the face that does not match the room
- Distortion when the head turns or a hand passes in front of the face
Audio signs
- Coughs or sneezes that do not match what is on screen. The FBI listed this as a reported cue.
- A flat or shifting voice quality
- Unusual pauses before every answer
Behaviour signs
- Reluctance to move, turn or change the camera angle
- Answers that sound read aloud or do not match the resume
- Resistance to any live follow-up step
What can you ask a candidate to do during the call?
Simple requests can expose weak fakes. Make them part of every interview so no one feels singled out.
- Ask the candidate to turn their head slowly from side to side.
- Ask them to pass a hand in front of their face.
- Ask about a specific line on their resume, then ask the same thing a different way.
- Ask them to share their screen and work through a small task live.
These tests will not catch every fake, and the tools behind deepfakes keep improving. Use them as one layer, not as the whole defence.
How do you verify candidate identity before and after the interview?
A layered process catches what a single video call misses. Here is a sequence that works for most hiring teams.
- Check the candidate profile before the first call. Look for consistent work history, email, phone number and professional links.
- Verify a government ID against the person, with consent. Use an identity verification step instead of a visual check on camera.
- Control the interview session. Use a platform your team manages, and record only with the candidate's consent.
- Add one live step. A screen-share task or a short working session is hard to fake in real time. For final rounds, consider an in-person meeting where it is practical. Gartner found that 62% of candidates were more likely to apply when in-person interviews were required.
- Verify references independently. Find the employer's contact details yourself. Do not use the number the candidate gave you.
- Match identity at onboarding. Confirm the person who starts work is the person you interviewed. Limit system access until identity is fully confirmed.
- Agree an escalation path. Decide in advance who pauses a call, who documents it and who reports it. In the US, that can include the FBI's IC3. In Canada, the Canadian Anti-Fraud Centre.
| Layer | What it helps catch | Limit |
|---|---|---|
| ID verification before the interview | Stolen or invented identities | Only as strong as the document check |
| Live checks on the call | Weak real-time fakes | Better fakes can pass |
| Live technical task | Stand-ins and scripted answers | Needs interviewer time |
| Detection software | Signs of manipulation in video or audio | False positives and misses |
| Independent reference checks | Invented work history | Slower |
| Onboarding identity match | Swaps after the offer | Happens late in the process |
Do deepfake detection tools work?
They help, with limits. Detection software analyses video, audio or both for signs of manipulation. It can flag a suspicious call for human review. It cannot prove anyone is real.
Expect false positives, which flag genuine candidates, and false negatives, which miss fakes. Attackers adapt, so tools that worked last year may lag. Some tools check only video or only audio. Results are probabilities, not verdicts.
Before you buy, ask any vendor:
- What does it analyse: video, audio or both?
- How does it handle false positives, and what happens to a candidate who is flagged?
- Where does candidate data go, and how long is it kept?
- Does it work inside your video platform and applicant tracking system?
- Who reviews a flag, and how fast?
What privacy and legal checks apply?
Recording interviews, running biometric analysis and checking ID all involve personal data. Tell candidates what you collect, why and how long you keep it.
Canadian employers should start with PIPEDA, the federal private-sector privacy law. For interview verification, it comes down to a few habits. Say why you are collecting data. Get the candidate's consent. Collect only what you need and keep it only as long as you need it. Protect it with security that fits how sensitive it is. Name one person who is accountable for compliance.
Provincial laws, such as Quebec's Law 25, may add requirements. US employers should check state privacy laws, including biometric laws such as Illinois' BIPA. Talk to legal counsel before adding facial recognition or other biometric tools. This is general information, not legal advice.
Candidate experience matters too. A flagged candidate may be genuine. Offer an alternative verification step instead of ending the process on one signal.
How does a verified talent marketplace fit in?
A check on a video call catches one kind of problem. Fake candidate profiles start earlier, with invented resumes, borrowed identities and skills that do not exist.
Every consultant on Augmnt is checked for resume fraud, skills, work eligibility and background. Businesses can find verified AI, data and tech consultants through the platform, or ask the Augmnt team to source, verify and deliver onshore consultants in the US and Canada.
You can still run your own interview checks. Verification at the source and checks on the call work best together.
Frequently asked questions
Can you detect a deepfake in a video interview without software?
Sometimes. Lip-sync problems, edge distortion and odd behaviour when the head turns can be clues. These cues are not reliable on their own. Pair them with ID verification and a live task.
What is the difference between a deepfake interview and a proxy interview?
In a deepfake interview, AI changes how a person looks or sounds. In a proxy interview, a different real person answers on the candidate's behalf. Both hide who is actually doing the talking, and the same verification steps help with both.
How common is candidate fraud?
In a Gartner survey of 3,000 candidates, 6% admitted to interview fraud. Gartner also forecasts that one in four candidate profiles worldwide will be fake by 2028.
Is it legal to record interviews and run facial analysis?
It depends on where you and the candidate are located. Get clear consent, explain how data is used and stored, and check Canadian and US privacy and biometric laws with counsel first.
What should I do if I suspect a deepfake candidate?
Do not accuse the candidate. Pause the process, ask for an additional verification step and document what you saw. Involve your security team. Report confirmed cases to the FBI's IC3 in the US or the Canadian Anti-Fraud Centre in Canada.
Which roles need the strongest identity checks?
Remote roles with access to systems, code, customer data or money. Software, data, AI, IT and finance roles are the usual examples.
Do in-person interviews solve the problem?
They remove many video risks. They do not stop stolen identities or invented work history, so keep the other layers in place.
Sources
- HR Dive, By 2028, 1 in 4 candidate profiles will be fake, Gartner predicts
- Gartner via Greenhouse, Mitigate Rising Candidate Fraud Through Identity Verification
- BleepingComputer, FBI: Stolen PII and deepfakes used to apply for remote tech jobs (FBI IC3 public service announcement I-062822-PSA, June 28, 2022)
- Penn Career Services, You are real now, prove it
Want to see how Augmnt verifies consultants? Request a demo.
