Half Your Applicants Might Be AI Fakes. Now What?

The Markup posted a single engineering job listing in late 2025 and received over 400 applications within 12 hours. Repeated phone numbers across different names. Fabricated companies that existed only on LinkedIn. Candidates who vanished the moment anyone asked a follow-up. That was one job post at one small publication.

We talk to hiring teams every week, and this pattern is everywhere now. The pipeline is not empty. It is full. Full of AI-generated resumes that mirror your job description word for word, take-home assessments completed by agents, and candidates who look perfect on paper but dissolve the moment you go off script. The problem in 2026 is not finding applicants. It is figuring out which ones are real.

91% of Recruiters Have Already Seen It

This is not a niche concern. A Greenhouse survey found that 91% of recruiters have encountered candidate deception, and a third said they spend up to half their week filtering junk applications. Gartner projects that by 2028, one in four candidate profiles globally will be fake.

From what we see on the ground, those numbers feel conservative. The teams we work with are not dealing with the occasional exaggerated resume. They are dealing with coordinated, AI-powered application campaigns where dozens of near-identical profiles hit the same role in a single day. The old signals that used to separate a real candidate from a fake one, a well-tailored resume, strong assessment scores, polished interview answers, are now the exact things AI generates best.

Adding More AI Filters Won’t Save You

The instinct is to fight AI with more AI. Better ATS filters. Automated fraud scoring. Checkr just launched an identity verification tool with liveness detection and forensic document analysis. Tools like that are genuinely useful as one layer of defense.

But if your entire strategy is more filters, you are just escalating the arms race. Candidates using AI to fake applications will adapt to your AI filters, because that is exactly what these tools do: optimize against whatever system they face. You cannot out-automate the automation. The teams actually winning are restructuring their process so the moments that matter most are human moments.

Camera On, Script Off, Judgment Over Output

Here is what we see working across the hiring teams we support.

Front-load verification. A quick camera-on call where a recruiter asks unscripted follow-ups about a specific project on the resume filters out most AI-assisted fraud immediately. Fraudulent candidates prepare for standard questions. They rarely survive improvisation.

Test judgment, not output. The old model was giving candidates a problem and checking if they produce the right code. The better model is giving them code that is almost right and seeing if they find what is wrong. You can AI your way through writing a function. You cannot AI your way through explaining why a design choice will cause a race condition under load.

Invest in pre-vetted talent over open pipelines. Referral hires and candidates sourced through trusted networks have dramatically lower fraud rates, faster ramp-up, and higher retention. When someone credible vouches for a person, that is a trust signal no AI can fabricate.

Fake Candidates Are Ruining It for Real Ones

This is the part that does not get talked about enough. When a third of your recruiting team’s week goes to filtering AI slop, they spend less time with genuine candidates. Those real candidates get a worse experience. Your offer acceptance rate drops. Time-to-fill goes up. The fakes are not just clogging your pipeline. They are degrading it for everyone in it.

And there is a subtler cost. When interviewers start assuming every polished application might be fake, they get more skeptical of everyone, including the excellent candidates who just happen to write well. Suspicion is corrosive. The best people have options, and they can tell when they are being treated like suspects.

Stop Praying the Inbound Pipeline Will Work

Add detection tools. Restructure interview loops. Train your team to spot red flags. All of that helps. But the single most effective move is to stop relying on open inbound applications as your primary talent source.

The companies hiring the best engineers right now are not posting and praying. They are working with partners who vet candidates before they enter the pipeline. Who verify skills, confirm identity, and validate work history before the first interview, not after the third.

That is what TechX does. We do not send resumes. We send people we have worked with, trained with, and verified. In a market where a quarter of your pipeline might be fiction, the most valuable thing we can offer is proof that a candidate is real, capable, and ready. If your team is spending more time detecting fakes than evaluating talent, let’s fix that.