How do I spot a proxy interviewer without accusing an honest candidate?

You do not accuse. You build verification into the process for everyone, so no single person feels targeted. Ask candidates to share their screen and walk through a live task. Ask a follow-up question that requires knowing what they just said two minutes ago. A proxy who is feeding answers to someone off-camera falls apart fast when the questions stop being predictable. Keep it standard for every candidate and it stops being personal.

Is candidate fraud really a staffing firm problem, or is it the client's problem?

It is your problem the moment you present the candidate. Your client hired you to vet people. If a synthetic identity or a deepfake gets through and lands on their payroll, you own the fallout. That means a clawed-back placement fee, a damaged relationship, and in some cases a contract clause about verification you did not meet. Treating fraud as a sourcing issue is how firms end up eating liability they never priced in.

What is the cheapest first step if I have no verification process at all?

Add a live, unscripted step to your final interview and require the camera on. That costs you nothing but a few minutes. Most fraud attempts collapse when a real person has to respond in real time to a question they could not prepare for. Start there this week, then layer in ID verification tools once you see how often profiles fail the live test.

The Candidate Fraud Playbook for Staffing Firms

Four in ten staffing buyers have already been hit by AI candidate fraud. Deepfakes, proxy interviewers, and synthetic identities are passing your ATS filters right now. Here is the verification playbook that protects your placements and your client contracts.

Four in ten staffing buyers have already dealt with an AI-driven fake candidate. That is not a prediction. That is where the market is today, and the number is climbing fast.

The fraud is not clumsy anymore. It is deepfake video in live interviews. It is a skilled proxy sitting off-camera feeding answers while a different face reads them back. It is fully synthetic identities, built by AI, that sail through your ATS keyword filters because they were written to match your req word for word.

Here is the part most firms miss. They treat this as a sourcing problem. Something the recruiter catches at the top of the funnel. It is not. It is a placement risk and a client liability problem that lands squarely on your P&L when a fake person clears your process and shows up on a client's payroll.

Why Your ATS Is Not Catching This

Your ATS was built to match resumes to job descriptions. It scores keywords. It ranks by fit. It was never built to ask one simple question: is this person real.

AI-generated profiles are written to win that keyword game. They pull the exact language from your posting and mirror it back. The synthetic candidate looks like the perfect fit because it was manufactured to look like the perfect fit.

So the profile scores high. The recruiter, moving fast on a hot req, sees a strong match and pushes it forward. Nobody in that chain is doing anything wrong. The tool did what it was built to do. The problem is that nothing in the workflow is checking identity.

That gap is where fraud lives. And it gets wider every quarter as the tools to fake a profile get cheaper and better.

The Three Fraud Types Hitting Staffing Firms Right Now

You need to know what you are actually up against. There are three main plays, and they each break at a different point.

The Synthetic Identity

This is a person who does not exist. AI builds the resume, the LinkedIn profile, sometimes a headshot and a work history that checks out on a surface glance. The identity is designed to pass your screening and get to an offer, often for remote contract work where nobody ever meets in person.

The tell: the paper trail is thin under the surface. No verifiable references who will pick up a phone. Employment dates that do not survive a real check. A digital footprint that is exactly as deep as it needs to be and no deeper.

The Proxy Interviewer

Here a real, unqualified person applies. But a skilled expert takes the interview for them, either on the call or feeding answers in real time. You hire the person who interviewed brilliantly. You get someone who cannot do the job on day one.

The tell: the candidate pauses before answering things they should know cold. Their eyes track something off-screen. Their answers get worse the second you go off the expected script.

The Deepfake

This is the newest and the scariest. Live video where the face on the screen is not the person behind the keyboard. The technology is good enough now that a rushed recruiter on a Friday afternoon will not catch it.

The tell: odd lighting around the edges of the face. Audio that lags the lip movement by a beat. Reluctance to do anything that breaks the frame, like turning the head fully sideways or holding up an ID next to their face.

The Verification Playbook

You do not fix this by buying one tool. You fix it by building verification into your process at the points where fraud actually breaks. Here is where to put the checks.

Live and unscripted at the final stage. Every candidate does a short live step where a real person asks questions they could not prepare for. Camera on, no exceptions. This alone kills most proxy and deepfake attempts because fraud cannot improvise well.
ID verification before the offer, not after. Use a tool that matches a government ID to a live selfie. Do it as a standard step for every finalist so no one feels singled out. If a profile refuses or fails, that is your answer.
References who pick up the phone. Not an email address. A real number, a real conversation, a real person who confirms the work history. Synthetic identities cannot produce this.
A skills task that matches the resume. Ten minutes of live work in the actual tool the job requires. A proxy who fed perfect answers in the interview cannot do the work when the expert is gone.
Consistency checks across the file. Does the resume timeline match LinkedIn? Do the dates survive a background check? Small gaps are normal. Manufactured histories have a specific kind of too-clean feel.

None of this slows you down much once it is built in. The live step takes a few minutes. ID verification runs in the background. The cost of adding it is tiny next to the cost of one clawed-back placement and a client who no longer trusts your vetting.

Why This Is a Client Liability Issue, Not a Sourcing One

When you present a candidate, you are vouching for that person. That is the whole reason a client pays your fee instead of hiring directly. Your promise is that the people you send are real and qualified.

A fake candidate that clears your process breaks that promise in a way the client will remember. You are looking at a refunded fee, a strained relationship, and increasingly a contract clause that spells out verification standards you may not be meeting today.

PE-backed portfolios should be paying extra attention here. Fraud that scales across multiple portfolio companies is a real exposure that shows up in diligence and in client churn. The firms that get ahead of this will use it as a selling point. Verified candidates, backed by a process you can explain in one page.

Do This Week

Pick your top three clients by revenue. Look at your last ten placements with each. Ask one question for every candidate: what step in our process would have caught a deepfake, a proxy, or a synthetic identity? Write down the honest answer.

If the answer for most of them is nothing, you have your priority for the quarter. Start by adding one live, unscripted, camera-on step to your final interview. That single change costs you nothing and closes the widest gap you have. Build the rest from there.

The Candidate Fraud Playbook: What Staffing Firms Must Do Before 1 in 4 Profiles Is Fake