Tendent

Candidate confessions

How candidates cheat on hiring tests, why they don't feel bad about it, and what changed by 2026. Quotes by role, methods, reasoning and the numbers.

Published on

September 28, 2026

We have written about the tools and about how to redesign the interview. This post is the other side of the table: what candidates actually do, how they explain it to themselves, and what has changed.

The quotes below are composites built from public forums and surveys, rewritten so nobody is identifiable. Only the role title is kept.

"The test had nothing to do with the job"

"The take-home was a CRUD API with pagination and filtering. I have built that thirty times. I pasted the brief into a chatbot, fixed two things it got wrong, and sent it back in forty minutes. I don't call that cheating. I call that a normal working day."

Backend developer, 8 years

The method: paste the brief, review the output, submit.

The reasoning: the task is generic, so the effort feels pointless. Candidates do not experience this as dishonesty. They experience it as not wasting a weekend on work nobody will read closely.

What it tells you: if a chatbot can finish your test alone, the test is measuring the chatbot.

"Everyone else is, so I'd be the only one who isn't"

"I ran a copilot on the side during the live coding round. It reads the screen and suggests the next step. Half my bootcamp group chat uses one. If I don't, I'm the only person in the pipeline solving it alone. That's not a level field."

Junior frontend developer

The method: an overlay tool that stays hidden from screen share and feeds hints in real time.

The reasoning: an arms race. Nobody wants to be the one who brought nothing. The tools market themselves as confidence aids, which makes the line easy to step over.

What it tells you: this is most common in junior and high-volume roles, where every candidate assumes every other candidate is doing it.

"They used AI on me first"

"My resume is rewritten by AI for every application. So is the cover letter. Then a robot reads it and rejects me in four seconds. At the one-way video stage I had the questions open in another tab and generated my answers before pressing record. They automated their side first."

Marketing manager

"Half my portfolio case studies were rewritten by AI so they sound like there was a process. The work is mine. The story isn't."

Product designer

The method: AI-tailored claims at every stage where a machine reads first.

The reasoning: reciprocity. If the company automates screening, candidates feel entitled to automate the answer. An HR consultant quoted by Newsweek put it bluntly: employers removed the people from their side of the process first, so candidates are answering in kind. Newsweek

What it tells you: claims-based stages (resume, cover letter, one-way video) are now noise on both sides. Most people are not lying about the skill. They are lying about the story around it.

"I use it all day at work, why not here?"

"I use AI all day. My manager expects it. Then the assessment says no AI tools. So which version of me are you hiring? I used it anyway and said so in the interview. Nobody knew what to do with that."

Data analyst

The method: work the way they work.

The reasoning: this is the one where the candidate has a point. Banning AI in the test creates a version of the job that no longer exists. The people building copilots know it: the CEO of Final Round AI says the goal is that "everybody can become Iron Man." yahoo

What it tells you: say what is allowed. Then measure how well they use it, instead of pretending they won't.

"I couldn't afford to fail"

"I was six months out of work. For the role-play round I had a script on a second monitor and a friend from my old team on mute in case I blanked. I'm not proud of it. I got the job. Three months in I was fine. The role-play was harder than the actual job."

Account executive

The method: scripts, a second screen, a person on a muted call.

The reasoning: pressure and volume. In the Resume Genius survey, 55% of job seekers named ghosting as their biggest frustration and 49% said the search has hurt their mental health. People who feel the process is stacked against them stop feeling bound by its rules. Newsweek

What it tells you: the last line is the important one. When the test is harder than the job, cheating on it looks rational.

The line most people won't cross

"Using AI is one thing. Paying someone to sit the interview as you is something else. That's not a shortcut, that's fraud. Most people I know draw the line there."

DevOps engineer

In a Gartner survey of 3,000 job seekers, 6% admitted to impersonation, either posing as someone else or having someone pose as them. That figure is a floor, not a ceiling. It is a small group with a different motive, usually money or access, not a shortcut through a tedious test. Candidates treat it as a separate category, and so should you. Alex

What changed between 2024 and 2026

Live is the new take-home. Two years ago the worry was the take-home. Now the copilot runs during the call. Greenhouse's 2026 research found recruiters observed live AI use in 35% of interviews, and 91% have spotted or suspected some form of deception. One interview-integrity dataset flagged 38.5% of candidates across 19,368 interviews, rising to 48% for technical roles. Thehirehub Metaintro
Admitted use is the floor. HackerRank reports only about 14% of candidates admit using generative AI on assessments, while 83% say they would if they believed it couldn't be detected. Resume Genius found 22% already use AI during live interviews, and nearly as many use it on skills tests. HeroHunt HR Executive
Technical roles lead. CodeSignal saw cheating on technical assessments climb from 16% to 35% of attempts in a single year, and Karat estimates 80% of candidates use LLMs on code tests even when explicitly banned. The Interview Guys
Assistance has normalised. Impersonation has not. Gartner predicts one in four candidate profiles worldwide will be fake by 2028. But the people doing that are a different population from the five above. Alex
Trust is broken in both directions. Only 26% of applicants trust that AI will evaluate them fairly. Candidates who think the process is unfair feel less obliged to play fair. HeroHunt
The bill arrives after the hire. Bloomberg Businessweek reported on a nonprofit that hired a grant writer after a polished interview, then found within a month that he froze on basic decisions; his manager is now almost certain he consulted a chatbot during the video calls. Bloomberg

What this tells hiring teams

Candidates cheat on the parts that feel disconnected from the job. Make the test the job.
Claims are cheap. Output is getting cheap. Understanding isn't.
Say what AI use is allowed, then measure how well they use it.
Ask them to explain the work. You can't defend work you didn't do.

This is why Tendent runs solve, then defend: candidates do real work, then walk an interviewer through their decisions. Nobody in the quotes above gets through the second half. Except the data analyst. That's the point.

Sources