AI and Your College Essay: What Every School's Policy Actually Means for You

Let's set the scene.

Your teenager has spent a Sunday afternoon staring at the blinking cursor of their Common App essay draft. At some point, a thought occurs: What if I just... asked ChatGPT for a little help?

Maybe just a sentence to get started. Maybe a quick outline. Maybe, just this once, a full draft to "get the ideas out" before rewriting it in their own words.

It's tempting. We get it.

Here's what you need to know before that happens — because the rules aren't uniform, they're not always fair, and the consequences range from "totally fine" to "your acceptance letter gets rescinded."

Welcome to the most confusing landscape in college admissions right now.

A robot reviewing a book

First, The Plot Twist Nobody Saw Coming

Before we break down what you can and can't do with AI, let's acknowledge the elephant in the room:

Colleges are using AI to read the very essays they're asking you to write by hand.

We're not making this up.

Virginia Tech now pairs every essay with one human reader and one AI reader. The AI can scan roughly 250,000 essays in under an hour — compared to two minutes per essay for a human.

That's not a typo. According to Virginia Tech's vice provost for enrollment management Juan Espinoza, the school is "saving at least 8,000 hours" of staff time per cycle and is now able to release decisions about a month earlier than before.

UNC Chapel Hill has been using AI to auto-score essays for grammar and writing style since 2019.

Caltech built an AI tool — a literal video "viva voce" exam — that grills applicants about their research projects on camera. An AI generates questions about your submitted work, you answer on video, and then faculty review the recording. It's essentially an AI-powered dissertation defense, and Caltech used it to screen roughly 10% of early applicants in its debut cycle.

Duke University went a different direction: it stopped scoring essays numerically altogether, citing "a rise in the use of generative AI" as a direct reason. Admissions offices can no longer assume, as Duke put it, that an essay accurately reflects the student's actual writing ability. So instead of scoring writing quality, they now read essays purely for personality and insight.

So: schools are using AI on essays while simultaneously telling students they can't use AI on essays.

We're not here to adjudicate that debate (though it's a good one). We're here to help you navigate the rules as they actually exist — school by school.

The Three Camps: Where Schools Stand Right Now

Camp 1: Full Ban. No Exceptions. We Are Watching.

BYU is the clearest example of zero tolerance. Their admissions page states explicitly: "You may not use generative AI tools (like ChatGPT) as you compose your responses." And then, right below that: "We use software tools to analyze the admission essays and activity responses of all students."

They mean it. Students whose essays are flagged as AI-generated may have their admission offer rescinded. There's also an attestation form — you sign it, you're on record.

Georgetown and Brown are in the same camp. Georgetown prohibits AI use in any portion of the application. Brown calls AI assistance "not permitted under any circumstances in conjunction with application content."

Yale calls it "application fraud" and has tied violations to potential expulsion.

Common App itself is firm on this too: using AI to generate application content is considered fraud across all member institutions.

The bottom line for these schools: Don't. Just don't. Not even a little.

Camp 2: Limited Use Is Okay — Within Boundaries

Several highly selective schools have taken a more nuanced approach, effectively saying AI is fine as a tool, but not as a ghostwriter.

• Caltech and Cornell: AI is permitted for brainstorming and grammar checks. Using it to outline or draft your essay crosses the line.

• Emory: AI can serve as a "coach" — think of it as a thought partner, not a co-author.

• WashU: AI for clarity and proofreading is fine. Idea generation or drafting is not.

If you think of it this way: using Grammarly is generally fine at most of these schools. Having ChatGPT write your first draft and then cleaning it up is not.

Camp 3: The Deliberately Ambiguous Middle

Here's where it gets murky — and where most schools actually live.

Stanford, Harvard, and MIT don't have explicit AI admissions policies on record. They require original work under existing honor codes, which technically covers AI misuse, but they've declined to publish specific guidance.

Wisconsin (UW-Madison) went the most transparent route of anyone: they explicitly state they will not disqualify an applicant found to have used or been suspected of using AI. They also confirm they don't run essays through AI detection tools. Their reasoning is essentially about fairness — they'd rather risk some AI essays than falsely accuse innocent students.

About 70% of schools have no published AI admissions policy at all. As in, complete silence. Which sounds like a green light but absolutely is not — most honor codes implicitly cover authenticity, and a flagged essay can still trigger a closer look.

What AI Detection Actually Does (And Doesn't Do)

Here's a sobering reality check for anyone thinking "the detection tools aren't that accurate anyway."

That's partially true — independent tests put most AI detectors at roughly 76% accuracy at best. There are false positives and false negatives. AI-edited human writing can slip through. Highly polished human writing can get flagged.

But here's the thing most students don't think about: a detection flag rarely ends the review right there. What it usually does is trigger a closer look — a human reviewer comparing your essay to your teacher recommendations, your graded writing samples, your short-answer responses. If your personal statement sounds like a TED talk and your activity descriptions read like a middle schooler wrote them, that inconsistency is noticed.

And some schools aren't using detection tools at all — they're relying on experienced admissions officers who have read thousands of essays and can recognize the patterns. AI writing tends to be grammatically immaculate, tonally consistent, thematically tidy, and emotionally... efficient. Human writing tends to be none of those things, and that's precisely what makes it human.

As Duke's admissions office put it, even as they moved away from scoring: "It's in reading those documents that the applicant comes alive to us."

An AI essay doesn't make an applicant come alive. That's the actual problem.

The Deeper Issue Nobody's Saying Out Loud

Here's what the data is really telling us, and it matters for how families approach this.

The essay has always been the one place in the application that was entirely yours. Grades, test scores, extracurriculars — all of those are years in the making and involve other people (teachers, coaches, tutors, programs). But the essay? That was supposed to be you, sitting with your thoughts, trying to say something true.

That's exactly why schools care so much about it — and exactly why AI disrupts the whole point of it.

When Duke eliminated numerical essay scores, they said the quiet part loud: they can no longer assume the essay reflects the student's actual voice. So instead of evaluating your writing, they now use essays purely to understand who you are. And guess what? An AI can't tell them who you are. It can only assemble a plausible version of someone.

Admissions officers are smart. They've been reading for years. They know the difference.

A Real Example of How This Can Go Wrong

Imagine a student — let's call her Maya — who is a genuinely strong writer. She had a fascinating topic: growing up between two cultures, navigating the tension between her immigrant parents' expectations and her own ambitions. Real, specific, emotionally layered.

She got nervous about the draft, pasted it into ChatGPT to "clean it up," accepted most of the suggestions, and submitted the revised version.

The result: a structurally cleaner essay that had lost every trace of Maya's actual voice. The specific, slightly awkward phrases that made it real — gone. The honest uncertainty in the ending — replaced with a tidy, motivational closing paragraph. The essay that survived sounded like a thoughtful essay written by no one in particular.

It didn't get her in anywhere she was reaching for.

Not because it was flagged. Because it was forgettable.

That's the real cost — and it has nothing to do with detection.

So What Should Students Actually Do?

Know your school's policy. Before you even think about using any AI tool, look up the specific policy for each school on your list. They're not all the same, and the stakes vary dramatically. A school-by-school check isn't optional anymore.

Use AI appropriately — and honestly — where permitted. At schools that allow it, using AI to brainstorm topics, check grammar, or research the college itself is fair game. Using it to write your essay — even partially — is where you cross a line that you can't uncross.

Remember what the essay is for. Admissions officers are reading your essay to find you. AI can produce a perfectly competent essay. It cannot produce your essay. Those are two different things.

Get real human feedback. Not AI feedback. Not just a parent reading it over. Feedback from someone who actually understands how admissions offices evaluate personal statements — someone who can tell you whether your voice is coming through and whether your story is landing the way you think it is.

That's Where My Admissions Sherpa Comes In

At My Admissions Sherpa, we work with students to write essays that are genuinely theirs — because those are the only essays that work.

We don't write essays for students. We help students find the story they actually want to tell, get their real voice on the page, and revise until the essay sounds unmistakably like them. We know what admissions officers are looking for when they read these things — including what makes them put an essay down after three sentences and what makes them sit up straighter.

In a cycle where AI essays are everywhere, the ones that land are the ones that couldn't have been written by anyone else. That's exactly what we help students create.

The rules around AI are changing constantly. Your student's story isn't. Let's make sure it gets told right.

→ Explore My Admissions Sherpa's Essay Support Services

Quick Reference: AI Essay Policies at a Glance

School	Student AI Use	School Uses AI?
BYU	Banned — rescission possible	Yes — detection software
Georgetown	Banned — all AI prohibited	Detection tools used
Brown	Banned — no exceptions	Not disclosed
Yale	Banned — "application fraud"	Not disclosed
Cornell	Limited — brainstorm/grammar only	Not disclosed
Caltech	Limited — brainstorm/grammar only	Yes — AI video "viva"
WashU	Limited — proofreading only	Not disclosed
Emory	Limited — coaching only	Not disclosed
Virginia Tech	Not formally disclosed	Yes — AI essay reader
UNC Chapel Hill	Not formally disclosed	Yes — since 2019
Duke	Not formally disclosed	No — eliminated essay scoring
UW-Madison	Permitted — won't penalize	No — explicitly stated
Stanford / Harvard / MIT	Implied original work required	Not disclosed

Policies change frequently. Always verify directly on each school's admissions website before applying.

Have questions about your student's essay strategy or which AI rules apply to their school list? Drop them in the comments — we're happy to help.