29% Rewrote Their CV for the Bot. 71% of Recruiters Don't Use One

If you've rewritten your CV to get past an AI screener, you're not alone: CV-Library's Candidate Behaviour Barometer found that 29% of UK candidates deliberately changed their CV or cover letter because they thought AI might screen it, almost three times the 2025 figure. In the same survey, 29% of recruiters said they use AI to screen all or some applications before a person sees them. So most candidates who rewrote for a machine are sending that CV to a recruiter who doesn't have one, and the rewrite can make it worse for the person who actually reads it.

Do recruiters use AI to screen resumes?

Most don't, at least at the CV stage. In CV-Library's August 2026 barometer (1,043 UK jobseekers and 233 recruiters, hiring managers and direct employers), only 29% of recruiters said they use AI to screen all or some applications before a human looks at them. An earlier CV-Library survey from March 2026 came in lower still: of the roughly 425 recruiters it asked, about 90 said they use AI at the CV screening stage, which is about 21%.

You'll see both numbers quoted, and people mix them up. The "1 in 5" figure is from March. The "29% vs 29%" comparison is from August. Whichever one you take, the conclusion holds: a clear majority of recruiters still have a human read the CV first.

Candidates assume the opposite:

Recruiters don't rate the tools highly either. In the same survey, early-stage screening (42%) and filtering CVs (41%) were among the tasks they said AI handles badly. It's hard to see why someone would hand the first read to a tool they think does it badly. That's the AI resume screening myth in a sentence: the fear is reasonable, but it's aimed at the wrong stage.

Who actually reads your CV at each stage?

Your application passes three stages, and the one most likely to reject you automatically is the knockout questions, not the CV parser. An ATS (applicant tracking system) is the database that stores, sorts and sometimes ranks applications. In most setups, a person still decides whether you move forward.

1. Knockout questions. Eligibility and compliance filters set by the employer. A wrong answer can reject you on the spot.
2. ATS storage and ranking. The system parses your CV into fields and may give it a fit score, which sets the order a recruiter reads in.
3. The human read. A recruiter or hiring manager opens the CV and decides.

The most-cited breakdown comes from Volen Vulkov's team at Enhancv, which surveyed 25 recruiters. Every one of them (100%) used knockout questions for compliance, and only 8% (2 of 25) had set up auto-rejection based on CV content. But 44% of the ATS platforms in that sample produced AI "fit scores" to rank resumes.

Treat the sample with care. Twenty-five recruiters is a small number, and the source sells resume tools. Alexander Troutner at the Zero Weeks Notice Substack makes a fair counterpoint: Enhancv's own data shows 44% of recruiters using ATS AI to rank candidates, and if your score puts you near the bottom of a large pile, you've effectively been rejected even though no software said no.

That leads to a more accurate rule than "the bot rejects you." Knockouts reject you. Fit scores bury you. Humans decide. Adecco CEO Denis Machuel has described the same split: his company uses AI in pre-screening while people make the final decision.

Because the hard rejections happen at the knockouts, answer eligibility questions exactly. Don't round up years of experience, and don't tick "willing to relocate" unless you mean it. That's a better use of your time than worrying about whether the parser can read a two-column layout.

Who is the human reader, really?

In the UK, the first person reading your CV is most likely an agency recruiter working at high volume, not a hiring manager reading slowly. In Refolk's index of professional profiles, 35,566 UK profiles have Recruiter, Talent Acquisition or Recruitment Consultant titles, and "Recruitment Consultant" is the most common of the three.

Very few of them describe themselves as AI-enabled. Only 111 of those UK recruiters, or 0.31%, put "AI" in their headline. The firms that come up most in that small group are Harnham, Forsyth Barnes, Gravitas and Matchtech, and many of those recruiters hire for AI roles rather than screening with AI. Recruiters who openly present themselves as running machine screening are rare.

For illustration only: if 21% to 29% of that UK pool used AI at the CV stage, that would be roughly 7,500 to 10,300 recruiters. The surveys weren't designed to represent this index, so read that as a rough sense of scale, not a measurement. Even at the high end, most of the people who read UK CVs are doing it with their own eyes.

That changes what you should optimise for. Agency consultants skim at volume, scanning for two things: a job title they recognise and a result with a number in it. Neither one is a keyword trick.

Why writing for the bot backfires with the human

Keyword-stuffing for a parser that probably isn't there gives the recruiter who definitely is there a reason to doubt you. In research reported by Personnel Today, 36% of recruiters said they were less confident than a year ago that a CV reflected a candidate's true skills. Among those, 52% blamed obviously AI-generated CVs or cover letters and 44% blamed exaggeration.

So the robotic rewrite isn't harmless. When you swap "cut onboarding time from two weeks to four days" for "leveraged cross-functional synergies to optimise onboarding workflows," you're producing exactly the kind of text that 52% of skeptical recruiters name as the reason they trust CVs less. Writing for an imagined bot sets off the filter in the real reader.

US research from Greenhouse in October 2026 (US respondents, n=1,200) shows how deep the loop goes:

• 77% of job seekers used AI in their search over the past year.
• 62% tried tactics to improve their chances, including using AI to work out what screening systems were looking for.
• 60% changed, or considered changing, their writing so it wouldn't look AI-generated.

That third figure includes people who only "considered" it, so it's softer than the headlines make it sound. Even so, the picture is candidates using AI to beat the screen and then trying to scrub the AI out of what they wrote. CV-Library's Alec Stanwell put it plainly: "Candidates are becoming so focused on writing for AI that they risk forgetting their real audience."

The response on the hiring side is to lean more on interviews and skills tests when CVs get harder to trust. The CV becomes a ticket to the next stage, and proof starts to count for more than polish.

Should you write your resume for the ATS or a human?

Write for the human, and stay out of the parser's way. The two goals rarely conflict once you separate them: the parser needs clean structure and clear facts, and the person needs specific evidence in plain language that doesn't read as generated.

What the ATS needs (do it once, then forget about it):

• Standard section headings: Experience, Education, Skills.
• Employer, job title and dates on their own lines.
• A plain file format the posting accepts.
• The posting's real requirements, named in its own words where they truly apply to you. If it names a specific tool or qualification, use that exact name, not a broader category.

What the human needs (this is where the effort goes):

• A title and top line that match the role you're applying for.
• Results with numbers: revenue, time saved, error rates, team size.
• Verbs that describe something that happened. "Spearheaded" and "leveraged" become "rebuilt the billing export" or "moved 40 clients to the new contract."
• Named tools, projects and team sizes, not categories.
• No adjective triplets. "Strategic, collaborative and results-driven" is the giveaway. Keep the one you can prove.
• Some unevenness. Real careers have one big win and several smaller ones. A CV where every bullet is equally impressive reads as generated.
• Nothing you'd have trouble saying to a hiring manager across a table. Read it aloud and rewrite anything you wouldn't say.
• A cover letter "why" you wrote yourself. Your reason for wanting this specific job is the one thing a model can't fill in convincingly.

Mirroring the posting's real requirements helps on both fronts. It gets you a better fit score where one exists, and it makes things easy for a recruiter skimming for a match. Stuffing in requirements you don't meet helps neither, and it's the exaggeration that 44% of skeptical recruiters say they've noticed.

Matching each posting without drifting into generic phrasing is tedious work, and it's what Refolk does for you. Paste in the posting and I rewrite your own resume for it from your real history and your own numbers, so it reads as your work rather than boilerplate, and I score how well you fit so you can see the gaps before a recruiter does.

The silent killer is usually the employer, not the software

If you're not hearing back, the more likely reason is a busy human or no reply at all, not a rejection by an algorithm. In Greenhouse's survey, 96% of candidates who got to interview had at least one complaint, and the most common was being ghosted by the employer (45%).

A lot of the silence you blame on AI is really someone who opened your CV, meant to get back to you, and never did. Rewriting your CV for a machine doesn't fix that. Sending more applications into the same silence doesn't either. Greenhouse CPO Sharawn Tipton summed up the mood: "For the first time I can remember, neither side thinks hiring is working."

Timing is something you can control. In one survey, 52% of recruiters said a resume submitted within the first 48 to 72 hours of a posting is significantly more likely to be reviewed. Where fit scores sort the pile and people read from the top down, applying early with a well-matched CV does more than any rewrite aimed at a bot.

If you think a fully automated decision did reject you, you may have options. People Like Us, a London non-profit, campaigns to make people aware of their legal right to challenge automated decisions. Ask the employer whether a person reviewed your application.

FAQ

Do recruiters use AI to screen resumes?

Some do, most don't. CV-Library's August 2026 barometer found 29% of UK recruiters use AI to screen all or some applications before human review, and its March 2026 survey found about 90 of roughly 425 (around 21%) use it at the CV stage. In Enhancv's breakdown of ATS setups, AI ranking was far more common than outright content-based rejection.

Does an ATS automatically reject my resume?

Usually not on content alone. In Enhancv's small sample of 25 recruiters, all of them used knockout questions but only 8% set up content-based auto-rejection. Still, 44% of the ATS platforms produced fit scores that decide reading order, so a poorly matched CV can sink without ever being formally rejected.

Should I write my resume for the ATS or a human?

Write for the human, and format so the ATS can parse it. Use standard headings, clear titles and dates, and the posting's real requirements in its own words. After that, every hour you spend should go into specific, numbered results a recruiter can check in seconds.

How do I make my resume not sound like AI?

Replace abstract verbs with concrete actions, name tools and team sizes, drop the lists of three adjectives, and read every line aloud. Of the 36% of recruiters who trust CVs less than a year ago, 52% blamed obviously AI-generated text, so specific details count for more now than polish.