By 2028, One in Four Job Candidates Will Be Fake. Small Teams Have the One Defense Money Can’t Buy.
For two years the worry was quality: is this candidate as good as the AI-polished résumé makes them look? The question has quietly changed to something more basic. Is this candidate a real person, and are they the person on the call? Gartner projects one in four profiles will be fake by 2028, and talent leaders now rank fraud as their top hiring challenge. The enterprise answer is an expensive detection stack. A small team has something cheaper and better.
For two years, the fear in a hiring inbox was quality. AI wrote the résumé, polished the cover letter, aced the take-home, so the question every founder learned to ask was whether the candidate was actually as good as they looked. That was a hard question. It was also the easy version of the problem. The question has quietly changed to something more basic and more unsettling: is this candidate a real person, and are they the person on the call?
That shift sounds like a fringe worry until you read the numbers. In a 2025 Gartner survey of 3,000 job candidates, 6% admitted to interview fraud, either posing as someone else or having someone else sit the interview for them. Gartner projects that by 2028, one in four candidate profiles worldwide will be fake. And in GoodTime’s 2026 Hiring Insights Report, a survey of more than 500 talent acquisition leaders, fraudulent or AI-assisted candidates ranked as the single most anticipated hiring challenge of the year, ahead of the perennial complaint about a shortage of qualified talent. The thing recruiters worry about most in 2026 is no longer finding good people. It is telling whether the person exists.
Are They Good, or Are They Real?
The reason this jumped from science fiction to a line item in a year is that the tooling got trivial. A convincing face swap and a cloned voice for a live video interview now take a single photo, a few seconds of audio, and consumer software anyone can download. The fully async remote pipeline, résumé to portal to one-way video to offer, was built for speed, and it turns out to be the perfect habitat for a synthetic candidate. Nowhere in that funnel does anything confirm that the applicant is a single, continuous, real human being. It was never designed to. It was designed to move fast, and speed with no human checkpoint is exactly what fraud needs.
A Fake Hire Is a Breach, Not a Weak Hire
A weak hire underperforms, and you manage them out. A fake hire is a different category of problem. In 2024 the U.S. Justice Department revealed that more than 300 U.S. companies, including Fortune 500 names, had unknowingly hired North Korean operatives for remote IT roles: workers using stolen American identities, routed through U.S.-based facilitators, funneling wages back to a sanctioned regime. One facilitator alone was sentenced to more than eight years for a scheme that generated roughly 17 million dollars. Separately, Checkr’s research found 23% of companies reported identity fraud among new hires.
Now scale that down to your team. The moment you onboard someone, you hand them a laptop, VPN credentials, and access to your code, your customer data, and your payroll. A large company has a security team, legal, and insurance to absorb it when that goes wrong. A ten-person startup hands the keys to whoever showed up and finds out later. The cost of getting identity wrong is not a missed quarter. It is your data, and if the wrong entity is on the other end of the offer, your legal exposure.
Why the Big-Company Fix Backfires
The enterprise answer is a defense stack: identity-verification vendors, document forensics, liveness detection, biometric checks bolted onto the top of the funnel. For a small team that is the wrong move twice over. First, you cannot afford it or run it, and the fraud tooling improves faster than the detectors, so you would be buying into an arms race you are structurally set up to lose. Second, and worse, it points suspicion at everyone. Every honest candidate gets treated as a probable fraud before they have said a word, asked to prove they are human through a gauntlet that makes your process feel like a border crossing.
We have made this argument before about AI-cheating surveillance and lockdown proctoring, and it holds here too. When you build hiring around catching the bad actor, you punish the good ones, and the good ones, the candidates with options, are the first to walk. Front-loading suspicion onto a pipeline that is 99% real people, to catch the 1% who are not, is a tax you pay on every genuine candidate. It is the losing move.
The Cheapest Defense Is a Live Human
Here is the quiet truth the defense-stack vendors do not advertise: synthetic candidates thrive in the absence of a real, unscripted human conversation. That is the one thing a small team has in abundance and a 5,000-person company has to manufacture. You do not need biometrics. You need a live, real-time conversation early in the process, camera on and off the script, where you ask the person to reason out loud about a specific decision on their own résumé or walk through their thinking on a real problem while you follow up on their answers. A face swap holds for a rehearsed monologue. It falls apart under a genuine back-and-forth with questions it could not prepare for.
Then separate two things people keep conflating. Assess the human continuously, through every conversation. Verify identity once, at the point of trust, at offer and onboarding, with a simple document and video confirmation rather than an interrogation at the application stage. You keep the funnel welcoming for the vast majority who are real, and you close the door at the exact moment it matters, when access is about to change hands.
The reason small teams get exposed at all is the same reason they are tempted to cut the human out of the top of the funnel: volume. When 400 applications land for one role, a live conversation with everyone is impossible, so the instinct is to automate the human away entirely and let an async pipeline sort it, which is precisely the environment fraud is built for. The fix is not more automated detection. It is getting the volume down to a shortlist small enough that a human can afford to talk to every finalist, because that live human is the defense no synthetic candidate survives. Scoring the flood against what the role actually needs, so the few worth a real conversation rise to the top, is the part we built Kynto to carry. It does not verify identity for you. It does something more useful: it makes the human conversation, the thing that actually catches the fake, affordable again.
Key Takeaways
- The hiring question shifted from quality to identity. Gartner projects one in four candidate profiles will be fake by 2028, 6% of candidates already admit to interview fraud, and talent leaders in GoodTime’s 2026 report rank fraudulent candidates as their top challenge of the year, ahead of talent shortage.
- For a small team, a fake hire is a security and legal breach, not just a weak one. The DOJ found more than 300 U.S. companies unknowingly hired North Korean operatives on stolen identities, and onboarding hands a stranger your data and access.
- Skip the enterprise detection stack and do not turn your funnel into a border crossing. Verify identity once at offer, and put a live, unscripted human conversation early in the process. That back-and-forth is what a deepfake cannot survive, and it is the one defense small teams already own.
The teams that handle this well over the next few years will not be the ones with the most expensive detection software. They will be the ones who kept a real human in the loop early, kept the funnel human enough that an impostor had to survive an actual conversation, and reserved that conversation for the few candidates worth it. Fraud is betting you will be too busy to talk to anyone. The move is to prove it wrong.
A live human conversation is the one thing a synthetic candidate cannot survive, but you can only afford it when the flood is already down to the few worth it. Kynto scores applications against what the role actually needs, so the real conversation you keep goes to the candidates who earn it.
See how Kynto scores the application flood