The CV mistakes AI cannot fix are the ones that depend on facts only you hold: what you really did, what it achieved, why there is a gap, how to reach you and what your own voice sounds like. AI can tidy grammar and structure in seconds. It can’t check your dates against your payslips, find your numbers or notice what your CV leaves out, and it will often invent details to make the page read better.

Recruiters are noticing. In a Robert Half survey of more than 2,000 US hiring managers, 65% said a surge in applications, many enhanced or generated by AI, has made it harder to verify candidates’ skills, and many have responded by spending longer reviewing applications and adding interview steps (Robert Half, March 2026). Against that background, a CV that is accurate, specific and clearly yours stands out.

This guide gives you eight checks to run yourself after the AI has done its part. For how to use AI well in the first place, see our guide to Career AI Tips for 2026.

Quick Answer

The eight CV checks AI can’t do for you:

  • Fabricated or inflated details: check every date, title, employer, qualification, tool and number against your records.
  • Missing metrics and scale: replace duty statements with numbers, scope and results.
  • A robotic, over-polished voice: write in your own words first and use AI only as an editor.
  • Over-designed layouts: use a clean, single-column, plain-text layout with standard headings.
  • Copy-pasted job descriptions: use the employer’s language only where an example backs it up.
  • Unexplained employment gaps: add one honest line for any significant break.
  • Wrong or risky contact details: use working, professional contact details and keep sensitive personal information off the CV.
  • A generic professional summary: open with proof of who you are and what you deliver.

AI can polish what you give it. These checks make sure what you give it is true, specific and complete.

Why AI Can’t Fix These Mistakes

An AI writing tool predicts plausible text from the information you give it. It has no access to your employment records, no way of knowing which achievements are real and no sense of what a recruiter in your industry will question. When information is missing, it tends to fill the gap with something that sounds right. When you paste in a job advert, it leans towards the advert’s wording. When you ask it to “improve” a CV, it drifts towards an average of the CVs it has seen.

That is why recruiters are reacting to AI’s typical output more than to AI use itself. In Resume Now’s 2025 survey of 925 US HR workers, 62% said AI-generated resumes without customisation are more likely to be rejected, yet 77% said they were more likely to interview a candidate who used AI thoughtfully to improve their resume (Resume Now, a CV-builder vendor). The target is generic writing, not AI.

Recruiters also read quickly. As explained in How Recruiters Screen CVs Quickly, they check relevance first and look at detail second. Each of the checks below targets a point where an AI-polished CV tends to fail that first read, or the follow-up questions that come after it.

1. Check for Fabricated or Inflated Details

The mistake: AI tools can invent or inflate metrics, project names, software and responsibilities to fit a job description. Robert Half’s release notes that generative AI tools can fabricate or embellish work history and skills, which is part of why verifying candidates has become harder.

Why AI can’t fix it: it doesn’t know your history. It writes what fits the page, not what happened, and it can’t tell the difference unless you do.

The check: audit every claim against your own records.

What to verify Check it against
Employers and dates Employment contracts, appointment letters, payslips or IRP5 tax certificates
Job titles The title in your contract or appointment letter
Qualifications Your certificates, and the institution that awarded them
Software, tools and systems Whether you could use or discuss each one in a conversation today
Numbers and results A report, KPI pack or email, or a former manager who could confirm them
LinkedIn profile The same titles, employers and dates as your CV (see what recruiters look for on LinkedIn)

Then apply a simple test: if you couldn’t explain a line for 30 seconds with a real example, change it or delete it. Early screening calls often probe exactly these details, which is one reason to read 10 More Recruitment Screening Questions to Prepare For before you apply. An exaggerated claim usually surfaces at interview, in reference checks or during verification, and that does far more damage than a modest but true CV.

2. Replace Vague Duties With Metrics and Scale

The mistake: a generic prompt produces generic bullets, such as “Supported process improvement across the department.”

Why AI can’t fix it: it can’t guess your volumes, budgets, team sizes or results. Unless you supply them, it either stays vague or makes them up, which brings you back to check 1.

The check: every recent role should carry at least one number or scope detail.

Vague
Supported process improvement and reporting.
Specific (illustrative)
Rebuilt the weekly stock-variance report for three depots, cutting month-end reconciliation from five days to two.

If you’re not sure where your numbers are, work through these five dimensions for each role:

Dimension Ask yourself Example (illustrative)
Volume How many, and how often? Processed about 60 supplier invoices a day
Time How long did it take before and after? Reduced onboarding from ten days to six
Money What budget, saving or revenue was involved? Managed an annual maintenance budget of R2.4 million
Quality What happened to errors, rework, safety or compliance? No lost-time incidents across 18 months
Scope How many people, sites, systems or clients? Supervised 12 artisans across two shifts

If you don’t have exact figures, use honest approximations (“approx. 40 accounts”) rather than precise-looking numbers you can’t defend. Not every job produces neat metrics, and scope and context still count. How to Prove Transferable Skills shows how to evidence skills when the numbers aren’t obvious.

3. Keep Your Own Voice

The mistake: uniform sentence lengths, predictable phrasing and buzzwords such as “dynamic”, “passionate” and “results-driven”, with every bullet following the same rhythm.

Why AI can’t fix it: AI writes towards an average of internet writing, not towards your voice. In the Resume Now survey above, 36% of respondents named generic content as a top reason for rejecting a resume.

The check: write your career history in your own words first, then use AI as an editor, not an author. A prompt that keeps it in that role:

Prompt to copy
“Act as an editor, not an author. Below is my CV text. Fix spelling, grammar and tense, and tighten long sentences. Do not add any facts, numbers, tools, job titles or achievements that are not already in the text. Keep my wording where it is clear. If something is vague, ask me a question instead of guessing.”

Then read the result aloud. If you wouldn’t say a sentence to a recruiter across a table, rewrite it in words you would.

4. Strip Back Over-Designed Layouts

The mistake: sidebars, skill bars, icons, text boxes, photos and charts, often with contact details tucked into the page header.

Why AI can’t fix it: AI produces text, not a file that an applicant tracking system (ATS) can read cleanly. Many CV builders look impressive to a person but present content in ways a parser can miss.

The check: look for the problems that genuinely cost you visibility:

  • Contact details only in the header or footer: many parsers skip those areas.
  • Text inside images or graphics: a parser can’t read it.
  • Non-standard section headings: content can be misfiled. Use plain labels such as Work Experience, Education and Skills.
  • Mixed date formats: they can distort your timeline or create false gaps. Pick one, such as “Mar 2022 – Aug 2024”, and use it throughout.
  • Skill bars and star ratings: they aren’t readable as text, and “4 out of 5” means nothing to a recruiter.

Two-column layouts are the most debated point. Some sources say modern systems handle them well, while older parsers struggled. Because you can’t know which system an employer uses, a clean single-column layout in a standard font such as Calibri or Arial removes the risk.

Myth check

An ATS rarely deletes your CV by itself. The often-quoted claim that 75% of CVs are rejected by ATS software is traced by ATS vendor HiringThing to a company that closed in 2013, and these tools mostly sort, rank and filter on criteria employers set, with people doing the reviewing. Formatting matters because it can hide your content from that ranking and search, not because a robot throws your CV away.

5. Don’t Copy-Paste the Job Description Into Your CV

The mistake: asking AI to make your CV “ATS-friendly” and getting back repeated keyword variations or whole phrases lifted from the advert.

Why AI can’t fix it: it matches words, not proof, and it doesn’t know which requirements you actually meet. Keywords do matter, because matching can be literal and near-synonyms may not count. But a CV that repeats the advert without evidence reads as hollow to the person who opens it.

The check: work through the advert in three steps.

  1. Underline the five to eight requirements that matter most.
  2. For each one you meet, use the employer’s term once, inside a bullet that shows where you used it.
  3. For each one you don’t meet, leave it off and be ready to address the gap honestly in your cover letter or interview.
Keyword stuffing
Stakeholder management, stakeholder engagement, stakeholder communication, stakeholder relations.
Keyword with evidence (illustrative)
Managed 12 site stakeholders, agreeing shutdown schedules with operations and contractors.

6. Explain Employment Gaps, Briefly

The mistake: leaving a silent blank between two roles.

Why AI can’t fix it: AI works with the text you give it and doesn’t notice what’s missing. A recruiter does, and a gap with no context invites assumptions. Harvard Business School research in 2021 found that around half of the participating US companies used filters that excluded candidates with a work-history gap of six months or more (Harvard Gazette). That is US data from 2021, and not every employer does this, but it shows why a short explanation matters.

The check: add one factual line for any significant break, dated in the same format as the rest of your CV:

  • Jan 2024 – Aug 2024: Full-time caregiver for a family member
  • Mar 2023 – Nov 2023: Completed [qualification or certification]
  • Jun 2025 – Oct 2025: Retrenched following company restructuring; upskilling in [area]
  • 2022 – 2023: Freelance [your field] work, described like any other role

Keep it short, neutral and factual, and don’t over-share personal details. If you’ve been retrenched, Resource Complete’s Retrenchment Help includes CV and interview coaching through its market readiness programme.

7. Fix Contact Details and Protect Personal Information

The mistake: dead links, a wrong phone number, an unprofessional email address from your school days, contact details hidden in the header, and too much personal information.

Why AI can’t fix it: AI formats whatever you give it. It doesn’t test your email or your links, and it won’t warn you about details that shouldn’t be on a CV.

The check: use this split, which follows common South African recruiter advice (Headhunters):

Include

  • A professional email, such as firstname.surname@
  • A number you answer, with a professional voicemail
  • A LinkedIn URL that opens your profile
  • City and province
  • Willingness to relocate, if relevant
  • Contact details in the main body, not only the header
Leave off until formally requested

  • Your ID number and copies of ID or passport
  • Date of birth
  • Marital status
  • Your full street address
  • Referees’ contact details (“References available on request”)

The reasoning is practical: identity theft is a real concern, and an employer that genuinely needs your ID number can ask for it later, once you’ve confirmed the employer is legitimate. Before you send, email your CV to yourself, click every link and call your own number.

8. Rewrite the Generic Professional Summary

The mistake: an opening line such as “A highly motivated professional seeking a challenging role to utilise my skills.”

Why AI can’t fix it: AI defaults to the phrases that appear on thousands of templates, because those phrases are safe. They say nothing specific about you.

The check: rewrite the summary as three or four lines using this formula:

[Job title] with [years] in [industry]. [Scope: what you manage or work on.] [One result, with a number.] Seeking [target role] in [location].

For example (illustrative):

Financial accountant with six years’ experience in manufacturing. Prepares monthly management accounts under IFRS for a three-company group and cut month-end close from eight working days to five. Seeking a senior accountant role in Johannesburg.

Match the first line to the title you’re applying for, and tailor it each time. The summary sits in the top third of page one, where a recruiter’s first read happens, a point developed further in The Future of Resumes and CVs.

Where AI Does Help: A Safer Workflow

None of this means avoiding AI. It is useful for the parts that don’t depend on facts only you hold:

  • Proofreading: spelling, grammar and consistent tense.
  • Tightening: shortening long sentences and removing repetition.
  • Structuring: turning your rough notes into clean bullets.
  • Digging out numbers: ask it, “Interview me with ten questions about this role that would help me find measurable results”, then answer from your own records.

Put together, a safe order of work looks like this:

  1. Gather your records and numbers first.
  2. Write a rough draft in your own words.
  3. Use AI as an editor, with the prompt above.
  4. Run the eight checks.
  5. Tailor the top third of page one to the specific advert.
  6. Read it aloud, and ask someone who knows your work to check the facts.

Once the CV earns a screening call, you’ll need to explain every line in your own words. Our guide to interview preparation with AI tools shows how to practise that without letting AI invent anything.

RC

Ready to Put a Stronger CV Forward?

Submit your CV to Resource Complete so recruiters can match you to suitable roles, or see what is open right now.

Submit Your CV

Browse current job listings

Frequently Asked Questions

Can AI write my CV for me?

AI can help you draft, edit and tighten a CV, but it can’t supply the facts. Your dates, titles, numbers and reasons for gaps have to come from you, and you need to verify everything it adds.

Do recruiters reject CVs written with AI?

Recruiters mostly react to generic or unverifiable content, not to AI use itself. In one US survey, 62% of HR workers said AI-generated resumes without customisation are more likely to be rejected, while 77% were more likely to interview candidates who used AI thoughtfully.

Does an ATS automatically reject CVs?

Usually not. Most applicant tracking systems parse, sort and rank applications using criteria the employer sets, and people review the results. Poor formatting can hide your content from that ranking, so keep the layout clean.

What should I leave off my CV in South Africa?

Many South African recruiters advise leaving off your ID number, copies of ID or passport, date of birth, marital status and full street address, and giving them only when formally requested. Use city and province instead.

How do I explain a gap on my CV?

Add one short, factual, dated line, such as caregiving, study, retrenchment or freelance work. Keep the tone neutral and avoid over-sharing. You can expand on it in an interview if asked.

How do I add numbers to my CV if my job isn’t measurable?

Use scope instead of results: how many people, sites, clients, systems or transactions you handled, how often, and over what budget or time period. Honest approximations are fine when exact figures aren’t available.

Sources and further reading: Robert Half: AI-generated applications survey (2026) · Resume Now: 2025 AI and the Applicant Report · Harvard Gazette: hidden workers study (2021) · HiringThing: ATS myths · Enhancv: ATS formatting mistakes · Headhunters: CV safety in South Africa. Survey figures are US data.