How do I spot an AI-generated fake candidate before the interview?

Watch the resume first. Fake profiles use perfect keyword matches with vague employment gaps and companies you cannot verify. Run a quick check on the phone number and email. Burner numbers and freshly created email domains are a red flag. If the LinkedIn profile has fewer than 20 connections and no real activity, slow down and verify.

Can deepfake candidates really pass a video interview?

Yes, and they are getting better. Real-time face-swap tools can hold up over a 30-minute call. The tell is behavior, not pixels. Ask the candidate to turn their head fully to one side, hold up a physical ID next to their face, or stand up and move. Deepfake tools break down with those requests. If they refuse or the video glitches on cue, end the call.

Whose job is fraud detection, the recruiter or the client?

It is yours. Your firm is the one placing the candidate, so your name is on the line if a fake gets through. Do not push verification onto the client. Build the checks into your process before submission. Clients who lose trust in your submissions do not come back.

AI Candidate Fraud Playbook for Staffing Firms 2026

GoodTime's 2026 report ranks AI-generated fake candidates above talent shortages as the top hiring challenge. Gartner says 1 in 4 profiles will be fake by 2028. Here is how to defend your pipeline this quarter.

Fake candidates are now the number one hiring challenge of 2026. GoodTime's 2026 Hiring Insights Report puts AI-generated fake applicants ahead of talent shortages. Read that again. The thing keeping staffing leaders up at night is no longer finding people. It is proving the people they find are real.

Gartner projects that 1 in 4 candidate profiles will be fake by 2028. That is not a far-off warning. That is a trend already showing up in your ATS right now. Every submission you send a client is a bet that the person on the other end is who they say they are.

The old defenses do not work anymore. A polished resume, a good LinkedIn, and a smooth video call used to mean you had a real candidate. Today all three can be generated in under an hour by someone who has never held the job they are applying for.

Why This Got Bad So Fast

Three things collided in the last 18 months.

First, remote and hybrid roles removed the in-person filter. Nobody shakes a hand anymore. The entire hiring process happens through screens, and screens can be faked.

Second, generative AI made fraud cheap. A fraudster can spin up a resume tailored to your exact job description, a matching LinkedIn profile, and a working reference script in minutes. The cost of creating a convincing fake dropped to almost nothing.

Third, real-time deepfake tools crossed the line from novelty to usable. Face-swap software now runs live during a video interview. The person you are talking to may not have the face they are showing you.

The motive is money. Fraudsters get a role, collect a few paychecks, sometimes steal data, then vanish. Some run the same trick across multiple firms at once using one skilled operator and several fake identities. Your firm eats the fallout when the client figures it out.

The Four-Layer Defense

You do not need a security team to fight this. You need a process. Here are the four layers that stop most fraud before it reaches a client.

Layer 1: Screen the paper

Before anyone talks to a candidate, the profile gets checked. This is fast and catches a large share of fakes.

Verify the phone number is not a VoIP or burner line. Free lookup tools flag these in seconds.
Check the email domain age. A domain created last week is a warning sign.
Look at the LinkedIn profile for real activity, mutual connections, and a history that predates the job hunt. Empty profiles with 12 connections are suspect.
Confirm at least one past employer exists and matches the dates on the resume.

Assign this to your sourcing team as a standard step. It adds about five minutes per candidate and stops the obvious fakes cold.

Layer 2: Test the live human

The video interview is where deepfakes get caught, if your recruiters know what to do. Train them on three simple moves.

Ask the candidate to turn their head fully to the left and right. Real-time face-swap tools distort at extreme angles.
Ask them to hold a physical ID next to their face for a beat. Fakes struggle to keep the swap stable during this.
Ask an unscripted, specific question about a project on their resume. AI-coached candidates freeze when the script runs out.

A candidate who refuses these requests or whose video conveniently glitches during them is telling you something. Believe them.

Layer 3: Verify the references for real

Fraudsters supply fake references with fake phone numbers. Do not call the number they give you. Find the reference independently through LinkedIn or the company directory, then reach out. It takes longer. It also closes the loophole fraudsters count on.

Layer 4: Document what you checked

Keep a short record of the verification steps you ran for each submitted candidate. If a client ever raises a concern, you can show your work. This protects your firm and builds trust with clients who are just as worried about this as you are.

What This Costs You If You Ignore It

One placed fraud can end a client relationship worth six figures a year. Picture the call. Your client hired someone you submitted, and now that person cannot do the job, or worse, stole data and disappeared. Your name is on that placement. The client does not blame the fraudster. They blame you.

Now multiply that. If 1 in 4 profiles is fake by 2028 and you are running no defenses, the math says you will submit fakes. It is not a question of if. It is a question of how many and how much it costs.

Firms that build fraud detection into their process turn it into a selling point. You can tell clients exactly how you verify candidates. That is a reason to choose you over the firm that still thinks a clean resume is proof of a real person.

Do This Before Friday

Pull your last 20 submitted candidates and run the Layer 1 checks on each one. Phone number, email domain, LinkedIn activity, one employer confirmed. It will take a couple hours. You will learn two things fast: whether any fakes already slipped through, and where your process has gaps.

Then write the four layers into a one-page checklist and give it to your team. Make it a required step before any candidate goes to a client. That single page is your firm's first real defense against the number one hiring problem of the year.

Fraud is not slowing down. Your process has to move first.

The AI Candidate Fraud Playbook: How Staffing Firms Defend Their Pipelines in 2026