Not reliably with software, but sometimes by reading. AI-detection tools make mistakes in both directions: they flag human writing as machine-made and can be evaded with small changes, so a detector score proves little either way.
What gives AI away to a recruiter is the writing itself: generic phrasing, claims with no evidence behind them, and details you cannot explain in an interview. Some employers also set rules, including asking you to certify that your answers are your own words. Read the instructions on every application, and make sure each line is true and yours.
Key takeaways
- AI detectors are unreliable. Stanford researchers found that they wrongly flagged most essays written by non-native English speakers, and OpenAI shut down its own detector because of poor accuracy.
- That cuts both ways. A detector cannot prove you used AI, and it can wrongly flag writing that is entirely yours.
- Recruiters notice content, not statistics: stock phrases, inflated claims and letters that could be anyone’s.
- Employer rules vary. Some prohibit AI in application materials, some ask you to certify your answers are your own, and many say nothing.
- The approach that holds up is your facts, your reasons and your own final edit, plus an honest answer if you are asked.
01
Can AI detectors tell if a resume or cover letter was written by AI?
Not dependably. One test of these tools comes from Stanford, where researchers ran essays through seven AI detectors. The detectors were near-perfect on essays written by U.S.-born eighth-graders, yet classified more than half of the TOEFL essays written by non-native English students, 61.22%, as AI-generated. Nearly all of those essays, 97%, were flagged by at least one detector. The researchers called the detectors unreliable and easily gamed.
MIT Sloan’s teaching technology group reaches the same conclusion in a guide titled “AI Detectors Don’t Work.” It says detection software has high error rates, and notes that OpenAI, the company behind ChatGPT, shut down its own AI detection software because of its poor accuracy. It also reports that simple editing of prompts reduced detection of genuinely AI-generated essays from 100% to 13%.
For a job seeker, three things follow.
- A detector result is not proof. An employer that relies on one will sometimes be wrong.
- Your own writing can be flagged. The risk is higher if English is not your first language.
- Passing a detector means nothing. Text that slips past software can still read as generic to a person.
Do employers run these tools at all? Some may. Yale’s Office of Career Strategy warns students that employers may scan materials to verify whether an AI tool was used, which may result in the denial of an application. Nobody publishes how many do, so assume it is possible and do not build your approach around it.
02
What actually gives AI use away?
A recruiter reading fifty applications for one role does not need software. The same signals come up again and again.
| Signal | What it looks like | Why it is noticed |
|---|---|---|
| Stock phrasing | “I am excited to apply,” “proven track record,” “passionate about” | The same sentences appear across many applications |
| No specifics | Skills and adjectives, with no project, number or name | Nothing in it could only have come from you |
| The posting handed back | Paragraphs that restate the job description as praise | It shows the input, not the candidate |
| Inflation | “Expert,” “led” and “spearheaded” on a junior resume | The claims do not match the titles and dates |
| Leftovers | A bracketed [Company Name], the wrong employer, a chatbot’s preamble | Nobody proofread it |
| Mismatch | A polished letter, then plain emails and a different story on LinkedIn | The documents do not sound like one person |
| The interview | You cannot expand on a line in your own resume | This is where invented detail ends an application |
Most of these are faults in any application. AI makes them easier to produce at speed, which is why a list of resume buzzwords to avoid is as useful after a chatbot as before one.
03
Do employers have rules about using AI in applications?
Some do, and they differ.
- Prohibition. Some application instructions say AI may not be used for part or all of the materials.
- Certification. Some ask you to confirm that the work is your own. The U.S. federal government is an example: its Merit Hiring Plan, issued in May 2025, adds four short essay questions to federal job announcements graded GS-05 or above, limits each answer to 200 words, and says applicants will be required to certify that they are using their own words and did not use a consultant or AI.
- Silence. Many employers say nothing about AI.
- Rules for later stages. Assessments and interviews often have their own instructions. What to expect is covered in AI interviews.
Harvard’s Mignone Center for Career Success gives the advice that applies everywhere: read all application instructions, because some organizations have rules about whether AI can be used in preparing application materials. Where there is a rule, follow it. Signing a statement that is untrue is a far bigger problem than a plainly written essay.
04
How do you use AI so the result is genuinely yours?
The aim is not to hide anything. It is to send an application that is accurate and specific, and that you can stand behind however it was drafted.
- Start from your own notes. Facts, results and your reasons for applying come from you, before any chatbot is involved.
- Use AI for structure and critique. Ask it to compare, question and tighten, as set out in how to use AI in your job search.
- Rewrite in your own words. Treat any generated sentence as a suggestion. For letters, the sequence in ChatGPT cover letter prompts is built around this.
- Add what only you know. One concrete detail in each paragraph.
- Check every fact. The risks of invented content are covered in should you use AI to write your resume?
- Keep your drafts. Notes and earlier versions show how the work developed if anyone asks.
- Skip “humanizer” tools and detector scores. Rewording text to get past unreliable software solves the wrong problem. Your reader is a person, and what that person checks is whether the content is true and specific.
- Answer honestly if asked. Use something like the answer below, adjusted to what you actually did.
Yes, as an editor. I wrote the content from my own experience, then used an AI tool to tighten the wording and compare it with the job description. I reviewed every line, and everything in it is accurate. I'm happy to go into detail on any part of it.
05
What if your own writing is flagged as AI?
It can happen, for the reasons the research shows. If an employer raises it, stay factual and offer evidence.
Hi [Name], Thank you for raising this with me directly. The [cover letter / answers] are my own work and describe my own experience. [If true: I used an AI tool only to check grammar.] I would be glad to walk you through any part of it, share my earlier drafts, or answer the same questions on a call. Best, [Your name]
Then be ready to do what you offered. A ten-minute conversation in which you explain your own examples settles the question better than any tool.
FAQ
Frequently asked questions
Can employers tell if you used ChatGPT on your resume?
Not reliably with software. Detection tools have high error rates, and resume bullets are short and formulaic whoever writes them. Recruiters do notice generic wording, inflated claims and details you cannot explain in an interview. A resume built from your real experience and edited by you gives them little to question.
Are AI detectors accurate?
No. In a Stanford study, seven detectors classified 61.22% of TOEFL essays written by non-native English students as AI-generated. MIT Sloan’s teaching technology group says detection software has high error rates, and notes that OpenAI shut down its own detector because of its poor accuracy.
Do employers use AI detectors on resumes and cover letters?
Some may. Yale’s career office warns that employers may scan materials to verify whether an AI tool was used, and that this may result in the denial of an application. There are no reliable figures on how common it is, and the tools themselves make frequent mistakes.
Is it dishonest to use AI on a job application?
Not when the facts are true, the final wording is yours and the employer has not prohibited it. It is dishonest to submit invented experience, or to certify that you did not use AI when you did. If the instructions rule AI out, write the materials yourself.
Can you be rejected for using AI on an application?
Yes, where an employer has a rule against it. Some application instructions prohibit AI or require you to certify the work is your own, as U.S. federal job announcements do for their essay questions. More often, applications are rejected for what AI tends to produce: generic or inaccurate content.
Should you use an AI humanizer to avoid detection?
No. Detectors are unreliable in both directions, so a low score proves nothing, and rewording generic text does not make it specific. Put the effort into the content: your own facts, a clear reason for applying, and wording you would be comfortable defending in an interview.
Sources
- Stanford Institute for Human-Centered AI: AI-Detectors Biased Against Non-Native English Writers: the study of seven detectors and the share of TOEFL essays wrongly classified as AI-generated.
- MIT Sloan Teaching and Learning Technologies: AI Detectors Don’t Work. Here’s What to Do Instead.: detector error rates, evasion by editing prompts, and OpenAI shutting down its own detector.
- U.S. Office of Personnel Management: Merit Hiring Plan: the essay questions on federal job announcements and the certification that answers are the applicant’s own words.
- Yale Office of Career Strategy: Using AI In Your Job Search: the warning that employers may scan materials for AI use.
- Harvard FAS Mignone Center for Career Success: AI for Resumes and Cover Letters: the advice to read application instructions for rules about AI.
Published by Jobbie and last updated on October 6, 2026. This guide is general information for job seekers, not legal, tax or financial advice. Spotted something wrong? Tell us.