Can ChatGPT generate a cover letter?

The AI Cover Letter Reality: Nearly Half of Job Seekers Are Already Doing It

Yes — ChatGPT can generate a cover letter, and there's a good chance the last one you received was written with it. According to survey data from Novorésumé, 46.8% of job seekers used AI to some extent on their most recent cover letter, and for 30.3% of them, AI did some or all of the actual writing.

Adoption isn't uniform, though — it skews young. The same survey found a clear generational gradient:

• Gen Z: 54.3% used AI on their last cover letter
• Millennials: 48.1%
• Gen X: 40.7%

That pace of adoption creates a genuine tension for job seekers. Cover letters still matter — 94% of hiring managers say they influence interview decisions, and nearly half say a strong letter could convince them to interview a candidate with a weaker resume. But when nearly everyone is using the same tool to write the same document, the letter's power as a signal starts to erode.

That erosion is measurable. A Yale-affiliated field study of more than five million cover letters on Freelancer.com found that after an AI writing tool launched, the correlation between letter tailoring and callbacks fell 51% — and the correlation with actually receiving an offer fell 79%. In other words, a polished, tailored letter no longer tells employers what it used to. Unsurprisingly, employers have responded by leaning on harder-to-fake signals, like past review scores, which grew 5% more predictive after the tool's arrival.

The volume problem compounds the signal problem. AI has accelerated applications across the board — U.S. applicants applied to roughly 15% more jobs in 2023 than the prior year, partly because AI made each application faster to produce.

So the real challenge isn't whether ChatGPT can write a cover letter. It can, in seconds. The challenge is that raw, unedited output now floods hiring pipelines — and hiring managers know it. Among AI-sensitive managers, 24% read obviously AI-generated text as low effort and 17% find it impersonal, per the Novorésumé data. The letters that still move the needle are the ones where AI handled the drafting and the human supplied the substance.

That's the workflow this article walks through: where ChatGPT genuinely helps, where it fabricates, and how to edit its drafts into something that survives a hiring manager's 30-second scan. It's the same principle we apply at SkillzBench when evaluating AI agent skills — the tool is only as good as the verification and judgment you layer on top of it.

The Hidden Risk: Why Unedited ChatGPT Output Can Cost You the Job

Here's the uncomfortable math: 88% of hiring managers claim they can spot AI text in job applications, and 39% of recruiters treat a blatantly AI-authored letter as a deal-breaker. Meanwhile, 29% of surveyed candidates say companies simply didn't hire them after discovering their ChatGPT use.

So does that mean you should avoid ChatGPT altogether? The evidence says no. As Novorésumé's research puts it, AI use itself isn't the dealbreaker — obvious, unedited AI text is.

The first danger is fabrication. ChatGPT is known to hallucinate, and it tends to invent skills and experiences that match a job description a little too perfectly. When your cover letter claims expertise your resume doesn't support, recruiters notice the mismatch — and any claim you can't defend in an interview can raise red flags and potentially cost you the job.

The second danger is the telltale language. Unedited AI output carries recognizable fingerprints:

• Overused words like "delve," "pivotal," "intricate," and "tapestry"
• Stock phrases such as "unwavering commitment" and "advocate for"
• Uniform sentence structure that reads as machine-polished
• Generic content that could belong to any applicant

That last point matters more than you might think. In a UNSW Business School classroom exercise, applicants given just one minute with ChatGPT produced letters that were nearly verbatim the same. A hiring manager reading their third identical letter won't blame the tool — they'll blame you.

The stakes are measurable. A Yale-affiliated field study of five million cover letters found that after an AI writing tool launched, the correlation between letter tailoring and callbacks fell 51% — cover letters became dramatically less informative as hiring signals. Employers responded by shifting toward harder-to-fake signals like past review scores. In other words, when everyone submits polished AI text, the text itself stops working in your favor.

The counter-evidence is just as clear. The same study found that a one-standard-deviation increase in editing time was associated with a 0.31 percentage point rise in offer probability — roughly 52% of the average offer rate among AI-assisted applications. Editing isn't optional polish; it's where the outcome is decided.

The pattern across every credible source is consistent: ChatGPT as a drafting assistant is fine, and even credited with real hires. ChatGPT as an unedited ghostwriter is where candidates lose offers. The tool isn't the risk — sending its raw output is.

That principle extends beyond cover letters. If you're building AI-assisted workflows for your job search — keyword extraction, tailoring, application automation — the same rule applies: verify what your tools produce before it reaches a human. SkillzBench is building a free directory of agent skills with security scanning and sandbox benchmarking, so you can see how a skill actually performs before trusting it with anything that matters. Join the waitlist to get access when it launches.

The Draft-Then-Edit Workflow: What the Research Says Actually Works

The difference between a generic AI letter and one that lands interviews often comes down to ten minutes. In a classroom exercise at UNSW Business School, applicants given just one minute with ChatGPT produced letters that were "nearly verbatim the same" across the group. Those given ten minutes, the actual job ad, and instructions to personalize the tone produced letters their evaluators described as noticeably stronger.

That time investment isn't just anecdotal. In a field study of AI-assisted applications on Freelancer.com, a one-standard-deviation increase in editing time was associated with a 0.31 percentage point rise in offer probability — roughly half the average offer rate among AI-assisted applications, according to the Yale-affiliated research. Editing is where the value gets created.

The UNSW researchers recommend escalating your prompts through four levels rather than asking for the whole letter at once:

• Start with a basic request for a draft.
• Add numbered criteria backed by evidence bullets from your own experience.
• Paste in company and role context, including the actual job description.
• Finish with instructions on personality and tone.

Students who worked "systematically and iteratively," paragraph by paragraph, produced stronger and more authentic letters than those who generated everything in a single prompt. The payoff is real: 13.4% of job seekers credit an AI-assisted cover letter with helping them land a job, and another 23.7% believe it might have, per Novorésumé survey data.

The key word is "assisted." ChatGPT can't research the company, verify your claims, or know which accomplishments actually matter — that substance has to come from you. Swap vague duty descriptions for quantified results, keep the letter under 400 words, and fact-check every claim, since AI tools tend to fabricate experiences that suspiciously mirror the job description. Only include things you can defend in an interview.

For readers building broader AI-assisted workflows around job applications — keyword extraction, tailoring at scale, ATS optimization — agent skills can automate those steps, and SkillzBench is building a free directory where each skill is security-scanned and benchmarked before you load it into your agent. The draft is the easy part; whether you join the waitlist or edit manually, the human pass is what hiring managers actually reward. As AI doesn't hire people — humans hire humans, and your editing time is what proves a human was there.

Your 5-Step ChatGPT Cover Letter Playbook

ChatGPT can write you a cover letter in ten seconds — but a UNSW Business School exercise found that one-minute, one-prompt letters came out "nearly verbatim the same" across every applicant. The good news: the same experiment showed that students who worked iteratively produced noticeably stronger, more authentic letters. Here's the five-step playbook that puts you in that second group.

Step 1: Feed it rich context. Paste the full job description and your actual resume into the chat before asking for anything. ChatGPT can extract keywords and mirror the employer's language, but only from what you give it — it can't research the company or invent your qualifications.

Step 2: Prompt iteratively, paragraph by paragraph. Follow the UNSW escalation: start with a basic request, then add numbered criteria with evidence from your resume, then company and role context, then tone instructions. Students who worked "systematically and iteratively" outperformed those who generated the whole letter in one shot.

Step 3: Fact-check every claim. ChatGPT hallucinates confidently, and it tends to fabricate skills that match the job description a little too perfectly. Mismatches between your letter and your resume dates are a documented red flag — include only claims you can defend in an interview.

Step 4: Add what AI can't supply:

• Quantified results — numbers and percentages, not day-to-day duties
• Company-specific research showing why *this* employer
• Brevity — keep the whole letter under 400 words

Editing pays off measurably: a Yale-affiliated field study found that a one-standard-deviation increase in editing time was associated with a 0.31 percentage point rise in offer probability.

Step 5: Rewrite the AI tells. Swap out "delve," "pivotal," "realm," and "unwavering commitment" for your own phrasing. This matters because obvious, unedited AI text is the real dealbreaker — 88% of hiring managers claim they can spot it, and 39% of recruiters treat a blatantly AI-authored letter as disqualifying.

If you want to automate the surrounding workflow — keyword extraction, ATS tailoring, drafting at scale — agent skills can handle those steps for tools like Claude Code, Cursor, or Gemini CLI. SkillzBench is building a free, independent directory where you can browse and test agent skills before loading them, with security scanning and sandbox benchmarking on the way. Everything — directory, MCP connector, benchmarks, alerts — will be free. Skills, proven. Join the waitlist at skillzbench.com to hear when it launches.