How can I tell if a candidate on a video interview is using a deepfake?

Watch for lag between audio and mouth movement, a face that stays unnaturally still, and lighting that never changes when the person moves. Ask them to turn their head fully sideways or wave a hand in front of their face. Most real-time deepfake tools break when the profile turns or something crosses the face. If they refuse to do it or the video 'glitches' right when you ask, that is your answer.

We are a smaller firm. Do we really need fraud detection tools, or is this just a big-firm problem?

You need it more, not less. Fraudsters target firms without process because they know the checks are thin. You do not need expensive software to start. Live verification steps, ID matching, and a documented interview protocol cost you nothing but discipline. One fraudulent placement that reaches a client can cost you the whole account. That math favors the small firm that tightens up now.

Won't strict fraud checks slow down our time-to-fill and annoy good candidates?

Done right, no. Real candidates expect verification and pass it in minutes. The steps that catch fraud, like a live ID check or a short unscripted question, add maybe five minutes to your process. Frame it as standard for every candidate and nobody blinks. The delay you should fear is the one that comes after a fake worker gets exposed on your client's payroll.

AI Candidate Fraud: The Staffing Detection Playbook

GoodTime ranks fake AI candidates as the #1 hiring challenge of 2026. Gartner says one in four profiles will be fake by 2028. Here is the detection playbook every staffing VP needs before the next placement goes sideways.

Fraudulent AI-generated candidates are now the number one hiring challenge of the year. GoodTime's 2026 Hiring Insights Report puts them ahead of the talent shortage itself. Read that again. Firms are more worried about fake people than a lack of real ones.

Gartner projects that one in four candidate profiles worldwide will be fake by 2028. That is not a distant problem. If you place contractors, you are already seeing it, whether you have named it or not.

The recruiter who submits a fake candidate to a client does not lose one placement. They lose the account. And in a PE-backed portfolio, one exposed ghost worker becomes a story that travels to every other firm in the group. This is a delivery risk problem now, not an IT curiosity.

What The Fraud Actually Looks Like

There are three flavors, and they show up differently in your pipeline.

Deepfakes. A real-time face and voice swap during the video interview. The person you are talking to is not the person who will show up. This is most common in remote tech and IT roles where the client never meets the worker in person.

Synthetic identities. A candidate built from stitched-together pieces. A real Social Security number paired with a fake name, a stolen resume, an AI-generated headshot, and a fabricated work history. Every piece checks out on its own. The person does not exist.

Ghost workers. One skilled operator interviews and wins the job, then hands the actual work to someone cheaper and less qualified overseas. The name on the timesheet never touches the keyboard. Your client pays senior rates for junior output.

The common thread is that all three pass a lazy screen. A resume review and a quick call will not catch any of them. That is exactly why they work.

The Detection Playbook

You do not need a six-figure fraud platform to start. You need a protocol that runs on every candidate, every time. Here is the sequence that catches the most fraud for the least effort.

1. Force a live liveness check

During the video interview, ask the candidate to do something a deepfake cannot fake in real time. Turn their head fully to the side. Wave a hand slowly in front of their face. Stand up and sit back down. Hold their ID next to their face and read a random number you say out loud.

Real-time deepfake tools still break when the profile turns past a certain angle or when an object crosses the face. If the video stutters at that exact moment, or the candidate suddenly has a technical problem, you have found your fraud.

2. Ask one unscripted question

Synthetic candidates and ghost proxies rely on scripts and prepared answers. Break the script. Ask something specific and human that no script covers. "Walk me through the last bug that ruined your afternoon." "What is the office coffee situation at your current job."

Watch for a long pause, a generic non-answer, or eyes drifting to read something off screen. A real person with real experience answers fast and messy. A fake stalls.

3. Verify identity against the document, not the resume

Match the government ID to the face on screen. Not later. In the interview. Free and low-cost ID verification tools now do this in under a minute, and they flag documents that were AI-generated or edited. Do this before you ever submit, not after the client asks questions.

4. Cross-check the digital footprint

Real people leave a trail. A LinkedIn profile created three weeks ago with 12 connections and a suspiciously perfect headshot is a red flag. Reverse-image search the photo. AI headshots often show up on multiple profiles with different names. Check whether the phone number and email tie to a real history or were created yesterday.

5. Watch the payment and location signals

Ghost workers reveal themselves after placement. A worker who insists on a payment method that hides their location, or whose login times never match their stated time zone, is telling you something. Build a check-in step in the first two weeks where the actual worker joins a quick video call with the client team. The proxy will not want to show up.

Build The Protocol Into Your Process, Not Around It

The firms getting burned are the ones treating fraud detection as a special case. They run it when something feels off. By then the fake is already in front of the client.

Make it standard. Every candidate for every remote or contract role goes through the same five steps. When it is universal, no good candidate feels singled out, and no recruiter can skip it to hit a number. This is a Build. Change. Adopt. problem. You build the protocol, you change the workflow, and you make sure the team actually runs it under pressure.

Document who ran each check and when. If a fraud slips through despite the protocol, you have a record that shows the client you have controls. That record is the difference between an awkward conversation and a lost account.

Train your recruiters on what fraud looks like on their screen. Most of them have never seen a deepfake break in real time. Show them a video of one failing. Once they see it, they cannot unsee it, and they catch the next one in seconds.

This week: Pick your five highest-volume recruiters and write down the exact verification steps they run today on a remote candidate. If the list has fewer than three steps, you have a gap. Add the live liveness check and the unscripted question to their next five interviews and see what surfaces. That is a one-day change that closes your biggest exposure.

The AI Candidate Fraud Playbook: Catching Deepfakes and Ghost Workers Before They Reach Your Client