You paste a job ad into a chatbot, tidy up the output, send the CV, and then the doubt arrives: can recruiters detect an AI resume? Sometimes they suspect it, they can rarely prove it, and most of them aren't looking for proof anyway. What hurts candidates is a document that could belong to anyone, and that problem costs interviews whoever wrote the words.
Suspected often, proven rarely
Recruiters read a lot of applications, and fast reading trains the eye for sameness. Writing assistants are now part of daily life, and that sameness has picked up a recognisable flavour: smooth sentences, confident verbs, nothing anyone could check. So yes, an experienced reader may suspect a CV had help.
That suspicion rarely turns into "this is AI, reject it". The usual reaction is the one described in what only you can write in a cover letter: this could have gone to anyone. A recruiter with forty applications and one afternoon has no reason to investigate authorship. They only need a reason to open the next file, and vagueness supplies one.
Help with a CV is also nothing new. People have long leaned on a friend, a career coach or a paid writer, and most hiring teams care far less about who polished the wording than about whether the content is true and relevant.
What a reader actually notices on a CV
Most talk about AI detection centres on cover letters, but a CV gives itself away differently. There is less prose, so the clues sit in the bullets and the summary.
- Bullets that only describe a mood. "Leveraged cross-functional collaboration to drive operational excellence" could sit on a logistics CV or a hospital CV. No product, no team, no outcome.
- A rhythm that never breaks. Eight bullets of the same length and the same build: strong verb, abstract noun, vague benefit. Real careers are lumpier, and some jobs deserve six lines while others get one.
- Phrases lifted from the ad. Echoing the employer's language is sensible. Mirroring it clause by clause, internal jargon included, reads like a mail merge.
- Skills with no trace in the history. The summary claims board-level stakeholder management, and the jobs underneath show two years as a junior analyst.
- Facts the model filled in. Ask a model to "make it stronger" and it may invent a team size or a result. This one is the dangerous clue, because it crosses from bland to false.
Take Lena, an invented account manager. Her AI draft says she "spearheaded strategic client relationships to maximise revenue growth". Her own version says she looked after 31 accounts at a regional print supplier and won back two that had left for a cheaper rival. Nobody else could have written the second line, and that is the whole difference.
What detector tools claim and what they can do
AI detectors are sold as a way to score how likely a text is to be machine-written. What they measure is statistical: how predictable the word choices are, how uniform the sentences look. A result of that kind is an estimate about style. It doesn't tell anyone who typed the text.
That creates problems for job applications. Carefully edited prose can look predictable. Candidates writing in a second language often use simpler, more standard phrasing, which is a recognised concern with these tools. A CV is harder still, since short fragments and stock phrases leave little to analyse, and the genre was formulaic long before chatbots.
An AI detection cover letter check has more text to work with, but the same weakness applies, which is why our cover letter myths piece calls these tools unreliable. Treat accuracy claims with caution. Paying to run your own letter through a checker and rewriting it to please the score is usually a poor trade, because you end up optimising for a test the recruiter may not be running and losing the specifics they do read.
Some hiring platforms may offer detection features, and that can change quickly. Nobody outside an employer can say what it runs, so write for a sceptical human, because one will read it eventually.
Employer policies vary, so read them
There is no single rule. Many employers say nothing at all. Others ask you to confirm in the application that the work is your own, or set limits on AI in written tasks and assessments. Graduate schemes, public bodies and regulated sectors are more likely to spell this out, though practice differs by country and organisation.
Read the instructions before you submit. If they forbid AI help, the issue is no longer detection: a box you tick is a statement you make, and breaking it can give the employer grounds to withdraw from the process. Take-home tests and writing samples deserve extra care, since their whole purpose is to see your own work.
Rules on how employers themselves use AI in hiring also differ from place to place, so check what applies where you work.
The real risk sits in the interview
An inflated line can survive the CV stage and collapse in the room. Say your CV reports that you "reduced churn through a data-led retention framework". A hiring manager only has to ask which customers, what you changed, how you measured it and who disagreed. If the honest answer is that a chatbot liked the phrase, the next ten minutes will feel long.
No detector scores this risk, and any decent interviewer will find it. Reference checks can do the same for titles, dates and team sizes. A polished claim you can't explain does more damage than a plain one you can defend for five minutes.
Run one test on every bullet: could you talk about it for two minutes without notes? Rewrite or cut whatever fails.
How to use AI without the tells
Using AI is a reasonable choice. For letters, when to use it and when to rewrite covers the decision, so here is the CV side.
- Start from your facts. List employers, projects, figures and names before you open a chatbot. Postulit works in a similar spirit by turning your LinkedIn profile into a CV, so the draft rests on your actual history and not on a blank prompt.
- Ask for edits, never for achievements. Good requests sound like "shorten this", "which bullet is weakest" and "which claims have no number". Avoid "make this sound more impressive".
- Add one detail per bullet that only you know. A client type, a budget, a deadline, the tool you replaced.
- Check every fact against your records. Dates, titles and figures need to match your contracts, your payslips and your LinkedIn profile.
- Read it aloud. Anything you wouldn't say to a colleague over coffee goes.
Before you send
Take the three strongest bullets on your CV and the opening paragraph of your letter. Add a specific figure, name or result to each, delete any claim you can't explain in two minutes, and re-read the employer's instructions on AI. Then send it, and keep a copy with your interview notes, because every line on it is a question you may be asked.