Summary
A cover letter used to signal effort simply by existing. Now that an AI cover letter generator produces one in seconds, existence signals nothing — Wharton research found that as AI spread, generic letters stopped carrying information, leaving specificity as the only remaining signal. Recruiters are not detecting AI; blind testing found 82% could not reliably identify it. They are detecting the absence of anything only you could have written.
Almost every guide to using an AI cover letter generator misses the thing that decides the outcome.
A cover letter never persuaded anyone by existing. It worked because writing one took forty minutes, and spending forty minutes was itself the message: this person wants this job specifically.
An AI cover letter generator removed that cost. Which means the letter no longer proves anything by being there — and the entire question of how to use one follows from that.
Do recruiters detect AI cover letters?
Not reliably. In a blind test of 1,000 recruiters, 82% could not correctly identify which letters were AI-generated, despite 74% claiming in surveys that they can. What they detect is the absence of specifics, which generic human-written letters share.
In practice, the gap between the claim and the test is the most useful evidence in this whole subject.
What recruiters say: a 2025 CV Genius survey found 74% believe they can identify an AI-generated cover letter. A separate survey put it at 88%, with 54% saying they would care.
What happens under test: ResumeBuilder.com ran a blind study with 1,000 recruiters, mixing human-written and ChatGPT-written letters without labelling them. 82% failed to correctly identify all the AI letters.
Both findings are real, and the reconciliation is the same one that applies to AI-written resumes: recruiters are not detecting authorship. They are detecting genericness — and a bad human letter fails the same test.
That has a measurable cost. Indeed’s 2024 research found hiring managers spend under 30 seconds on obviously AI-generated letters against two to three minutes on authentic-sounding ones. Not a rejection, exactly. Just four times less attention.
Why did the cover letter stop signalling anything?
Because effort became free. Wharton's Judd Kessler found that as AI writing tools spread, generic letters stopped carrying information — the act of submitting one no longer proved anything, since anyone could produce one in seconds. Personalisation is what remains, because it still costs something.
Fundamentally, this is the economics underneath the whole debate, and it explains the advice better than any list of tips.
A signal only works if it is expensive to fake. A cover letter was once a costly signal: producing a decent one required reading the posting, researching the company, and writing for half an hour. Nobody did that for a job they did not want.
Now the floor is a button. So the population of letters includes everyone, and the letter’s mere presence tells a recruiter nothing about your interest.
What survived
What still costs something is specificity. A model cannot invent that you rebuilt the onboarding flow at Wexford and cut activation time from eleven days to three. It cannot know that the company shipped a feature last month that solves a problem you have lived with. Those details are the only part of the letter that survived the collapse.
A field study on AI writing assistance in applications makes the same point over time: callbacks rose at first, then faded as employers adapted. The advantage was temporary because it was available to everyone.
The test for every sentence
Read each sentence and ask whether another candidate applying to a different company could have written it. "I am passionate about delivering results in a fast-paced environment" passes for anybody. "Your November post about deprecating the legacy pipeline is the exact migration I ran at Wexford" could only come from one person. Delete every sentence in the first category — you will usually cut half the letter, and it will be better.
Should you write one at all?
It depends on the company and the route in. Survey data genuinely conflicts — 72% of recruiters say they read letters when submitted, while one 2026 survey found 51.7% never do. The pattern is that smaller employers, senior roles, referrals and applications with under 25 candidates are where letters get read.
Notably, the conflicting numbers are not a contradiction; they measure different populations.
| Situation | Does it get read? |
|---|---|
| Large enterprise, hundreds of applicants, optional field | Rarely, and often not until late |
| Small company or startup | Usually, and often first |
| Role with under 25 applicants | Usually |
| Career change or employment gap | Yes — it is the only place to explain |
| Referral or direct route to a hiring manager | Yes |
| Explicitly requested in the posting | Yes, and skipping it is a real risk |
That last row deserves weight. A Saddleback College study found nearly half of hiring managers have passed on an otherwise-qualified candidate who omitted a requested cover letter. When the posting asks, the letter is a compliance test before it is a persuasion document.
The reverse is also true. LinkedIn Easy Apply and Indeed Quick Apply have no cover letter field at all, and for a mass application to a large employer with an optional field, a generic letter adds nothing and costs you time you could spend on a better-targeted application.
How do you use an AI cover letter generator properly?
Constrain it. A thin prompt produces the average of every cover letter ever written. Supply the full job description, your actual resume, real numbers from your work, and one specific fact about the company — then edit hard. The tool handles structure; you supply everything that makes it yours.
With an AI cover letter generator the failure mode is always the same: too little input.
1. Paste the whole posting. Not the title. The full description, including the parts about team structure and what the first months involve.
2. Paste your real resume. With the numbers in it.
3. Add raw notes. Two or three things you did that relate to this role, in plain language, with figures you can defend.
4. Give it one company-specific fact. A product change, a public post, something a scrape would not surface.
5. Forbid invention. State explicitly that it must not add a number, tool or employer you did not supply.
6. Then edit for five minutes. Cut every sentence that would fit another application. That edit is where the letter becomes yours.
Which model an AI cover letter generator runs barely matters. Testing across ten generators found drafts from newer and older models were indistinguishable — what varies is the input, not the engine.
Write your raw material first and let the tool compress it. You supply the facts; it handles the phrasing and the structure.
Paste a job title and press generate. With nothing of yours to work from, the output is necessarily the average of every letter — which is precisely the thing that stopped signalling anything.
How long should it be?
150 to 250 words across three tight paragraphs for most roles, up to 400 for senior or technical positions. Among letters that led to interviews in one 2026 sample, the median was 287 words — and letters over 400 words consistently underperformed.
Meanwhile, length is where good letters most often die, because the instinct when you have something to prove is to write more.
Recruiters skim the first two or three sentences, and the whole letter typically gets under 30 seconds. A 400-word essay is not read more thoroughly; it is read less.
Three paragraphs is the shape that works: why this specific role, one concrete proof point with a number, and a clean close. Front-load the fit, because the opening is the only part you can rely on being read.
Generic (could be anyone): "I am writing to express my interest in the Senior Analyst position at BrightPath. With over six years of experience in data analysis, I am confident I would be a valuable addition to your team."
Specific (could only be one person): "Your posting mentions the warehouse migration slipping twice. I ran the same sequence at Wexford in the wrong order, spent six weeks unpicking it, and would start by measuring which reports actually break before touching the schema."
The first opening is what an unconstrained generator produces. The second required thirty seconds of thought and one real memory — and it is the only one that earns the next paragraph.
What gives an AI letter away?
Opening with "I am writing to express my interest", achievements with no attributable numbers, uniform sentence rhythm, corporate vocabulary pitched above the role, and no reference to anything specific about the company. Each is a property of thin input rather than of the tool.
Five patterns give an AI cover letter generator away, and an edit fixes all of them:
The stock opening. “I am writing to express my interest in the position” is the single most common tell.
Unattributable achievements. Results with no numbers, no named systems, no dates.
Uniform rhythm. Every sentence the same length. Human writing varies.
Elevated vocabulary. Language pitched above the role reads as unedited output.
No company specifics. The clearest signal of all, and the easiest to fix — one genuine observation takes two minutes.
With the democratisation of AI, generic letters no longer signal anything. The effort of personalisation has become the only signal that matters.
Finally, the same logic runs through the rest of the application. A resume built from a job description alone has the same problem, which is why beating AI resume screening comes down to specificity rather than tooling — and an AI resume checker will show you whether the specifics you kept actually made it onto the page.
Frequently asked questions
Can recruiters tell if I used an AI cover letter generator?
Not reliably. A blind test of 1,000 recruiters found 82% could not correctly identify which letters were AI-generated, despite 74% claiming in surveys that they can spot them.
Will using AI get my cover letter rejected?
Not for using AI. What gets rejected is generic writing, which human-written letters produce too. Indeed's research found hiring managers spend under 30 seconds on obviously generic letters against two to three minutes on specific ones.
Do cover letters still matter in 2026?
Selectively. They matter most for smaller employers, senior roles, career changes, referrals, and postings with under 25 applicants. For mass applications to large employers with an optional field, they add little.
How long should an AI-generated cover letter be?
150 to 250 words across three paragraphs for most roles, up to 400 for senior or technical positions. In one 2026 sample of letters that led to interviews, the median was 287 words.
What is the best AI cover letter generator?
The differences between tools matter less than the input you give them. Testing across ten generators found drafts from different underlying models were indistinguishable — thin prompts produce generic letters regardless of the tool.
Should I skip the cover letter if it's optional?
For a mass application to a large employer, usually yes. When a posting explicitly requests one, do not skip it — nearly half of hiring managers have passed on a qualified candidate who omitted a requested letter.
How do I make an AI cover letter sound like me?
Supply real material before generating: the full posting, your resume with numbers, raw notes about relevant work, and one specific fact about the company. Then cut every sentence that would fit another application.
What is the biggest giveaway in an AI cover letter?
Opening with "I am writing to express my interest in the position", followed closely by achievements with no attributable numbers and no reference to anything specific about the company.
Comments