Safe AI help · Updated 2 September 2026

Can You Use AI on Your Personal Statement?

Application season has a rule most applicants never read: the certification box. Here is what schools actually permit, what counts as fraud, and the workflow that keeps AI useful without putting your offer at risk.

You can almost always use AI to think, often to proofread, and almost never to write. Submitting AI-generated content as your own is treated as application fraud rather than as a normal academic slip, and on most applications you personally certify that you did not do it. The specific rule varies by school — sometimes between two programs at the same university — so the first job is not writing, it is finding the certification text on the application in front of you and reading it.

The certification is the actual rule

Most applicants look for a page titled "AI policy." The binding text is usually somewhere less obvious: the statement you tick to submit. On the Common Application, and on many institutional applications that borrow its language, you affirm that everything submitted "is my own work, factually true, and honestly presented."

Brown University's master's application spells out what that means for AI, and it is worth reading in full because it is the clearest version of the standard most schools are converging on:

"While I may use artificial intelligence to assist with spelling and grammar review (in the same way as any other platform that supports basic proofreading), the content of all essays, short-answer questions and any other material submitted by me is my own work. Furthermore, I acknowledge that it is considered application misrepresentation to submit plagiarized essays or other written or oral material, or intentionally misrepresenting as one's own original work: (1) another person's thoughts, language, ideas, expressions, or experiences or (2) the substantive content or output of an artificial intelligence (AI) platform, technology, or algorithm (including but not limited to ChatGPT, Gemini, Bard, and Llama, for example)."

Two things follow. First, the line is drawn at substantive content, not at whether a tool was open in another tab. Second, this is a statement you sign, which is why the consequences are unusually heavy — Brown's application integrity page lists "admission revocation, expulsion, or revocation of course credit, grades, and degree" among the outcomes for misrepresentation. That clock does not stop when you enrol.

The rules are not uniform — even inside one university

Brown is the clearest illustration of why "what's the policy on AI?" has no single answer. Its master's certification permits AI proofreading. Its undergraduate admission page states that "the use of artificial intelligence by an applicant is not permitted under any circumstances in conjunction with application content." Same institution, opposite defaults, different application.

Note the direction of travel in Penn's August 2026 guidance: the test is not stylistic, it is whether you can account for your own sentences out loud. That is the standard to write toward regardless of which school you are applying to.

Law school is the strictest corner

If you are applying to JD programs, treat generative AI as off-limits for drafting unless a specific school says otherwise in writing. Two reasons.

First, the stakes extend past admissions. Many law schools ask applicants to certify their materials were prepared without generative AI, and Northwestern's prelaw guidance warns that a violation "can result in a Character and Fitness issue for state bar admission purposes" — a problem that surfaces three or more years later, when it is far more expensive.

Second, the written rules lag the expectations. In Kaplan's 2024 survey of admissions officers at 97 law schools, 45% said their school had an official policy prohibiting applicants from using generative AI to write the essay, 1% had a policy allowing it, and 54% had no official policy at all. For brainstorming, 16% allowed and 16% banned, with 68% silent; for feedback on a self-written draft, 15% allowed, 15% banned, 70% silent. Most applicants are operating in that silence — and silence is not permission, because the certification still binds.

What about detectors?

Less than the internet suggests, and the industry says so itself. LSAC has written that the development of AI tools is outpacing detection tools, pointing to OpenAI withdrawing its own detector over accuracy problems. LSAC also tested the obvious idea — compare your proctored LSAT Writing sample against your personal statement and flag mismatches — and concluded the technology "is not yet accurate enough to be integrated into LSAC's current offerings."

Do not read that as a green light. It means the real check is human and conversational: an interview question, a mismatch between your essay's voice and your graded work, an inability to explain your own argument. It also means honest applicants get flagged by imperfect tools, which is a separate problem — if that happens to you, our guide on what to do when you're falsely accused of using AI covers the evidence that actually clears you. Keeping dated drafts, as Northwestern advises, protects you in both directions.

UK applicants: the format changed

If you are applying through UCAS for 2026 entry onwards, the free-text personal statement is gone. It has been replaced by three structured questions — why you want to study the subject, how your studies have prepared you, and what you have done outside education — inside the same overall 4,000-character limit, with a 350-character minimum per answer. Practically, this makes generic AI output easier to spot, not harder: the questions ask for specifics only you have.

Where AI genuinely helps

The useful work happens before and after the draft, never during it.

• Getting material out of your head. Most weak statements fail because the applicant could not remember what they actually did. Interrogation beats generation here.
• Structural critique of a draft you wrote. Where the argument sags, what a reader still doesn't know about you, which paragraph is doing no work.
• Checking against the prompt. Programs ask specific questions; applicants answer the question they wish had been asked.
• Proofreading, where permitted, as the last step and nothing more.

Four prompts that stay on the right side of the line

1. Read the actual rule before you write. Paste in the certification text and instructions from your application.

Below is the certification text and application instructions for a program I am
applying to.

Tell me, quoting the exact phrases:
- what it says about AI, if anything
- whether proofreading is addressed
- what I am certifying when I submit
- what it does NOT address, so I know where I am guessing

If it is silent on AI, say so plainly rather than inferring a rule.

2. Interview yourself for raw material. This is the highest-value prompt on the page, and it writes nothing.

Act as an admissions interviewer for a [subject] [master's/PhD/JD] program.

Ask me one question at a time about my experience, motivation and specific
moments that shaped my interest. Dig into details: what I actually did, what
went wrong, what I decided. Wait for my answer before the next question.

Do not write any part of a statement. After 12 questions, give me a plain
bulleted list of the concrete specifics I gave you, in my own words, so I can
draft from it.

3. Critique the draft you wrote — without rewriting it.

Here is a personal statement draft I wrote, and the prompt it must answer.

Do not rewrite it or suggest replacement sentences. Tell me:
- which parts do not answer the prompt
- which claims about me are stated but not evidenced with a specific example
- what a reader still does not know about me after reading it
- which paragraph is weakest and why

Quote the exact phrases you are flagging.

4. The explain-it-back test. Penn's standard, run as a check before you submit.

Here is my finished statement. Ask me five questions that an interviewer would
ask based on it - the kind that check whether I actually lived what I wrote.

After I answer, tell me where my spoken answers and my written statement do not
match in detail or in voice.

If AI feedback starts flattening how you sound, keeping your own voice when AI gives feedback covers how to take structural notes without absorbing the tool's register. The same discipline applies to job applications — see tailoring a resume and cover letter with AI.

If your school allows AI with disclosure

A small number of programs permit AI use if you declare it — a model that is likely to spread. Where that option exists, take it and be specific: name the tool, the stage, and what it did. "Used Claude for grammar review of my final draft" is a complete disclosure. Vague disclosure is worse than none, because it invites the reader to imagine the maximum. Our AI disclosure statement guide has wording you can adapt, and if you are unsure whether to ask at all, the email template for asking about AI use works on admissions offices too.

FAQ

Can I use AI to write my grad school personal statement?

Submitting AI-generated content as your own is prohibited almost everywhere, and at many schools you actively certify that you did not do it. Brown's master's application makes applicants agree that it is application misrepresentation to submit as one's own original work "the substantive content or output of an artificial intelligence (AI) platform," naming ChatGPT, Gemini, Bard and Llama. Using AI to help you think, or to proofread text you wrote, is treated very differently from using it to generate the essay.

Is using AI for spelling and grammar allowed?

Often yes, but not universally. Brown's master's certification explicitly permits AI "to assist with spelling and grammar review (in the same way as any other platform that supports basic proofreading)." Penn's August 2026 guidance groups using AI to organize thoughts or check grammar with acceptable help like a teacher proofreading. But Brown's undergraduate page says AI is "not permitted under any circumstances in conjunction with application content." Two policies at one university — read the one in front of you.

Do admissions offices run personal statements through AI detectors?

Some do, but the sector is openly sceptical. LSAC has written that AI development is outpacing detection, noting OpenAI withdrew its own detector over accuracy. LSAC also tested comparing the proctored LSAT Writing sample against personal statements and concluded the technology "is not yet accurate enough." The practical risk is less a detector score than a certification you signed and an interview question you can't answer in your own words.

What happens if a school decides my essay was AI-written?

It's handled as application fraud, and the consequences can arrive long after you enrol. Brown lists "admission revocation, expulsion, or revocation of course credit, grades, and degree" among possible outcomes for misrepresentation. Law applicants carry an extra layer: a false certification can become a character and fitness issue at state bar admission years later.

Do law schools have clearer AI rules?

Stricter in tone, still patchy in writing. In Kaplan's 2024 survey of 97 law schools, 45% had a policy prohibiting generative AI for writing the essay, 1% allowed it, and 54% had no official policy at all. For brainstorming and for feedback on a self-written draft, roughly 68-70% had no policy either way. Silence is not permission — the certification still binds.

Bottom line

Find the certification text before you write a word, and let it set your defaults rather than guessing from a blog post. Use AI to interrogate yourself and to critique what you wrote; never to produce the sentences you submit. Keep every dated draft. Then apply Penn's test honestly: if you couldn't explain how you arrived at a line, in your own words, in a room with someone, it isn't ready to send.