39% Were Wrongly Flagged as Fakes. Make Your Work History Checkable

If a recruiter thinks you're a fake candidate, arguing rarely fixes it. What works is making your history so easy to check that the question answers itself.

A lot of honest people are in that spot. On October 14, 2026, Textio and Kyle & Co published an interview report with two numbers that don't fit together. Talent leaders rated their interviewers 8.1 out of 10 at catching live AI help. Meanwhile, 39% of job seekers said they had been wrongly suspected of cheating or misrepresenting themselves.

Why are so many honest candidates being flagged as fakes?

Employers are hunting a rare problem with tools that make mistakes, and in that setup most alarms go off on real people. How confident interviewers feel doesn't change the arithmetic.

This is the base-rate problem. A false positive is an honest candidate the system marks as suspicious. Here are the two inputs:

• In a Gartner survey from 2Q25, 6% of candidates admitted to interview fraud, either posing as someone else or having someone pose as them.
• Independent testing puts AI detectors' false-positive rate on human-written technical text at roughly 12% to 26%.

Picture 100 candidates, and say 6 are frauds, matching the share who admitted it. If a detector misfires on 12% to 26% of the other 94, it flags roughly 11 to 24 honest people. That's true even if it catches every fraud. Text detectors are only one tool, but the logic carries over: when the real thing is rare, false flags outnumber true ones.

The report also shows how differently the two sides see the same process. Employers scored their own interview process at +48.25 on a standard recommendation scale. Candidates scored the process they had just been through at +2.87. One disclosure: Textio sells interview intelligence software, so it benefits from that gap looking large. Even so, a gap of 45 points isn't a rounding error.

What are candidate fraud detection tools actually looking for?

Most fraud detection does two things. It compares your sources against each other, and it watches for unusual behavior on camera. It isn't judging whether you're honest. It's looking for things that don't line up.

A hiring-integrity tool is software that checks whether a candidate is who they say they are and whether they're getting outside help. In its June 2026 fraud report, Kyle & Co counted:

• 60+ vendors claiming a fraud or hiring-integrity capability
• 9 stages of the hiring funnel, each with its own attack surface
• 0 vendor-neutral response playbooks

Employers didn't buy these tools out of paranoia. The cases are real:

• KnowBe4 hired a North Korean operative who passed every hiring control it had.
• Pindrop, itself a fraud vendor, caught a live deepfake face overlay in one of its own interviews. The same identity came back eight days later through a different recruiter.
• Nisos followed one suspect to a 40-device laptop farm.
• Socure reports that a quarter of its own applicants are fake.

Gartner's often-quoted claim that 1 in 4 candidate profiles could be fake by 2028 is a forecast, not a measurement.

Because there's no shared playbook, a flag at any of those nine stages can end your process without anyone asking you a follow-up question. Matt Charney of Kyle & Co put it this way: "Eighty-five percent of employers already record interviews, and candidates still can't tell you how they were judged." From your side, a flag feels like silence. The recruiter stops replying, the next round never gets booked, and you never learn which mismatch caused it. So you keep applying into a system that has quietly decided you're a risk.

The signals that make a real candidate look fake, and the fix for each

Five signals cause most of the damage: dates that don't match, vague employers, generic bullets, glances off camera, and choppy video. You can fix each one before anyone asks.

Resume and LinkedIn dates don't match

To a fraud tool, a mismatch looks the same whether it comes from fraud or from a profile you haven't updated. Refolk's index of professional profiles shows how common that drift is, even among real people:

The title-match share comes from a 25-profile sample per query, so read it as a direction, not a precise figure. The pattern still matters. Between 36% and 40% of the AE and India engineer samples don't currently hold the title they match on. Their headline, past roles, and current title tell different stories. Some of those people got promoted, some moved on, and some just never cleaned up their profile. A cross-source check can't tell which.

The fix is boring and it works:

1. Pick one canonical version of every role: employer's legal or common name, exact title, and start and end month and year.
2. Use that version everywhere: resume, LinkedIn, application forms, and background check paperwork.
3. If your title changed at the same company, list both titles with their own dates rather than only the latest one.
4. Make your LinkedIn headline describe the job you have now.

The hard part is keeping that canonical record intact while you tailor your resume for every posting. Refolk is built for exactly that. It writes your resume from one master history, so you can rewrite the bullets for each posting without the dates or employer names quietly drifting between versions.

Vague employers and founder titles

Real careers include entries that are hard to verify. In Refolk's index sample, legitimate US engineer profiles list "Self-Employed" as an employer and carry Founder or Co-Founder titles. In the India sample, 7 of the 25 people matching "Software Engineer" currently list a Founder or C-level title.

Those entries are honest, but nothing can check them. Give a reviewer something to check: the product name, a live URL, a GitHub org, a client you can name, or a former collaborator who will vouch for you. "Founder, stealth startup" invites suspicion. The same line with the product's name, what it does, and a link to it invites a question about the product instead.

Why engineers start out under suspicion

Software engineers face the highest bar because employers have been told to expect fraud in their applicant pools before a single interview happens. Fabric, a vendor, reports a 48% AI-cheating rate in software engineering interviews versus 12% in sales. Gartner advises employers to assume at least half of applications for remote IT positions are false.

Put that next to the index numbers. Refolk's index has 353,813 US Software Engineer profiles and 248,612 US Account Executive profiles. The engineer pool is 1.42x the size of the AE pool, but engineers are flagged at 4x the rate. India's engineer pool, at 582,678 profiles, is 1.65x the US pool.

The result: an honest engineer applying to a remote role begins the process already doubted. Offering proof early is a smart move, not an admission of anything:

• Offer an in-person round if the company has an office you can reach.
• Volunteer to do ID verification before the technical screen.
• Link public work the reviewer can check without logging in.

Gartner's own survey found 62% of candidates are more likely to apply when a company requires in-person interviews. Asking to meet in person puts you with the majority.

How to prove identity in a remote interview

To prove who you are on video, remove the technical glitches that look like fraud and say out loud what you're doing. The behaviors that get honest people flagged are ordinary ones: a nervous candidate who looks away to think, a person with ADHD who fidgets, a shaky connection that makes video look choppy enough to trigger a deepfake alert.

Before the call:

• Use a wired connection or sit next to the router. Lag is the cheapest signal to remove.
• Test your camera and lighting well before the call, not five minutes before it.
• Close extra monitors and overlays. Recruiters have tools like Cluely, a live-answer overlay, in mind when they watch your eyes.
• Ask for accommodations early if fidgeting or breaking eye contact is part of how you think.

During the call:

• Say it when you look away. "I'm checking the notes I wrote on this project" turns a suspicious glance into an ordinary one.
• Think out loud. Talk through tradeoffs, name the systems involved, and give the numbers. Interviewers are trained to treat polished answers with no specifics as possible AI output.
• Pause on purpose, and keep it short. "Give me ten seconds to think" is fine.

If you're accused of using AI in an interview, offer to do a follow-up on the spot: share your screen, walk through a past project in detail, or come in person. Someone using a live-answer tool can't easily go deeper on a project they never did. You can.

Which LinkedIn verification actually counts?

LinkedIn's identity badge is the weakest proof you can get, and colleague-confirmed work history is the strongest. The badge only shows that a verification partner checked a government ID and that LinkedIn matched it to the name on the profile. It doesn't show that your jobs, schools, or photo are real.

The badge is still worth having. LinkedIn reports that verified members get 60% more profile views. But the badge answers "is this a real name?" Fraud tools are asking "is this a real work history?"

That's why the pilot LinkedIn announced in September 2026, where connections confirm your work history, matters more. It covers the claim fraud tools actually test. If you're an engineer, or anyone in a heavily flagged field, ask two or three former teammates to confirm your roles now, while you're calm and not under suspicion.

What to do if a recruiter thinks you're a fake candidate

Answer quickly with evidence, not indignation. A recruiter who flagged you is usually following a process with no playbook (Kyle & Co counts zero vendor-neutral ones), so make it easy for them to clear you.

1. Ask which detail raised the concern. Dates, employer, video, or answers each need a different fix.
2. Send a one-page timeline of every role with exact dates, titles, and a reference or public link for each.
3. Fix any mismatch the same day. If LinkedIn and your resume disagree, correct whichever one is wrong and tell them you did.
4. Offer stronger proof: ID verification, a reference call, or an in-person round.
5. Keep it short and factual. A long defensive message looks worse than a short, checkable one.

Then fix the root cause in your canonical record before the next application, so the same mismatch doesn't follow you to another employer.

FAQ

Can a fraud tool reject me just because my resume and LinkedIn dates don't match?

Yes. Many hiring-integrity tools compare sources against each other, and a mismatch looks the same whether it's fraud or an old profile. In Refolk's index sample, 36% to 40% of people matching Account Executive or India Software Engineer don't currently hold that title, so drift is common. Make dates, titles, and employer names match exactly across every document.

Does a LinkedIn verification badge prove I'm a real candidate?

Only partly. The badge confirms that a government ID matched the name on your profile. It says nothing about your jobs, schools, or photo. Colleague confirmation of your work history, which LinkedIn began piloting in September 2026, covers what fraud tools actually check.

What should I do if I'm accused of using AI in an interview?

Offer to go deeper right away. Share your screen, walk through a past project in detail, or propose an in-person round. AI detectors misfire on 12% to 26% of human-written technical text, so a flag isn't proof. Answering with specifics shows you have experience a live-answer tool can't fake.

Is it suspicious to list Self-Employed or Founder on my resume?

Not by itself. Real engineer profiles in Refolk's index include both. The problem is that neither can be checked. Add the product name, a live link, a client you can name, or a collaborator who will confirm the work, so a reviewer can verify the entry in a minute.