Across five Clear Admit MBA Essay Insight webinars this July, admissions leaders from 23 business schools sat down to discuss their essay questions—and, at the beginning of every session, their school’s policy on applicants’ use of artificial intelligence in the admissions process. Representatives from schools including Chicago Booth, Columbia Business School, Georgetown McDonough, INSEAD, Northwestern Kellogg, SDA Bocconi, SMU Cox, Washington Foster, and Wharton all addressed the topic.
Their policies were worded differently.
Their underlying message was not.
If you are applying this cycle, here is the reality, drawn directly from what these officers said on the record.
Nobody Is Telling You Not to Touch AI
Almost no one on the panels told candidates to avoid AI entirely. The admissions officers use AI themselves. They know you do, too.
“You can use any tool you like,” said J.R. McGrath of Carnegie Mellon Tepper. Columbia’s Kate Murphy commented, “We encourage and permit people to use AI in their process, whether it be for proofreading or to help with idea generation.”
Michigan Ross’s Jahnee Horn was blunt about why pretending otherwise is silly: “We absolutely understand that it’s a great tool. Many of us are using these tools in our day-to-day work.”
INSEAD’s Jennifer Liu put the obvious on the table, remarking, “I don’t think anyone is not using AI right now.”
So the question was never whether you can use AI. Every meaningful part of the guidance is about how far you can take it — and that line turned out to be remarkably consistent.
The One Rule Every School Shares: Assist, Don’t Author
Strip away the different policy language and you get a single dividing line repeated 23 times: Use AI to support your writing, never to produce it.
The permitted uses cluster tightly around proofreading, grammar, and, for some schools, brainstorming or structure:
- Berkeley Haas, Cindy Millett: “We’ve seen evidence over the last three cycles of written materials, even some scripts in our video essay, that are definitely AI generated. And in truth, it really does flatten out the uniqueness of each individual. When we’re reviewing applications… it should be written by you. However, we know people are using AI to brainstorm content or structure, not to create—in that way it is a really good tool.”
- Dartmouth Tuck, Pat Harrison: “We allow it for limited use for editing assistance, but it should not be generating complete responses, substantial content; it certainly should not be used to translate your answers into English.”
- Yale SOM, Kate Botelho: “We tell you to use it like you would use a trusted friend. So you would never have a friend write your essay, I hope, but you would use them to maybe proofread or help you kind of in brainstorming.”
- Johns Hopkins Carey, Jenn Howder: “It can be used in a similar way to any other platform like a Grammarly, or the same thing that a friend or a mentor might do if they’re revising your essay.”
The prohibited use is always the same: letting the tool generate the substance.
Duke Fuqua’s David Sieber drew the line even tighter than most, cautioning against leaning on it for ideas at all: “We do not want you to use it for content creation… preferably not even brainstorming.”
Kellogg’s Katie Dyssegard was direct: “Don’t take this time to just put our essay prompt into AI and submit that, because it’s doing you a disservice.”
The through-line word, said in nearly every session, was authenticity.
As Tuck’s Harrison summed it: “Your ideas, your voice, your experiences should all be your own… If a tool is creating your content, you sound like a computer. You don’t sound like you, and authenticity is really what’s going to get you admitted to all of our schools.”
Why AI-written Essays Quietly Sink You
The strongest argument against generating your essay isn’t that you’ll get caught.
It’s that AI makes you sound like everyone else.
Chicago Booth’s Nicole Chen explained the mechanism: “The more that you put into them, you’re not only going to sound less like yourself, you’re going to sound more like everybody else, because everyone’s using the exact same tools.” Her sharpest line has become something of a rallying cry: “We’re not admitting ChatGPT to the program.”
Cornell Johnson’s Hannah Harlow was even more direct: “I would probably discourage you from using AI to write your essays. They all look exactly the same, even if you think you’ve tweaked it to make it not look the same. AI cannot recreate your brain even if you give it information. So use your brain, skip the AI if you can.”
Georgia Tech Scheller’s Megan Friedman framed it as an old problem made worse: “Even before AI had such a boom, we would always tell people, we don’t want you to come across as a business robot. And now, even more so, it’s easy to come across as a business robot if you’re just using AI as your crutch.”
Remember the volume these readers work through. One admissions director cited on the panels estimated she’d read 56,000 applications over her career. Against that backdrop, generic writing doesn’t just fail to help — it actively buries you.
The Policies Vary More Than the Guidance Does
Here’s where applicants get tripped up: the guidance is near-universal, but the formal policy is not. Schools fall into a few different camps, and you are often signing an agreement whether you notice it or not.
Schools with an explicit AI statement in the application
Berkeley Haas
Chicago Booth
Duke Fuqua
Georgia Tech Scheller
Wharton
Northwestern Kellogg
SMU Cox
Yale SOM
Wharton’s Park Watson commented, “We do have a policy, but it really reaffirms the perspective that everything that you provide to us should be authentically from you.”
SMU Cox’s Janene Buttinelli-Bell noted candidates actively check a box: “We do have a statement that students check within their application saying that they will not use AI within their application” — even as, she stressed, the school itself embraces AI academically.
Schools that fold it into honesty or honor codes rather than a standalone AI rule
Georgetown McDonough
UVA Darden
UVA Darden’s Christian West said, “We don’t have an official policy on the use of AI in the application process. However, the University of Virginia is very well known for its academic honor code… We will teach you how to use these tools, but also do so responsibly. And we ask you to do the same.”
Georgetown McDonough requires a verification statement plus, distinctively, “a list of AI do-nots. We don’t give dos. We give do-nots to help people understand what’s off the table,” said Anne Kilby.
Schools with “no formal policy” but clear expectations
Emory Goizueta
UNC Kenan-Flagler
Johns Hopkins Carey
SDA Bocconi
INSEAD
London Business School
Cornell Johnson
These schools all said they have no formal rule — and then, in the same breath, described exactly the assist-don’t-author expectation. UNC’s Lica Hertz explained, “At this time, we don’t have a policy regarding AI usage in the application, but we encourage responsible, ethical usage.”
Schools that asks you to disclose
SDA Bocconi
Washington Foster
Washington Foster stands slightly apart. “Our policy is that you can use it however you’d like, but we want you to tell us how you used it,” said Megan Lewis. “It should not write the essay for you.”
SDA Bocconi’s Alessandra Gazzola similarly “strongly recommend[s] candidates to disclose if they use AI, but to not use AI to craft their answers.”
The practical takeaway, echoed by Clear Admit’s host Graham across the sessions: Read the fine print for every school on your list. Some have a specific AI statement; for others, it’s baked into the terms you agree to when you submit. Don’t assume — check.
Some Readers Scan for It. All of Them Can Feel It.
A few schools are candid that detection is part of the process. Duke Fuqua’s Sieber said, “We do scan for — it can get pinged if it feels overly AI in our system,” while acknowledging the technology will keep evolving on both sides. Berkeley Haas has seen it firsthand across cycles, including, per Millette, “some scripts in responses to our video essay” that were clearly AI-generated.
But most officers didn’t lean on detection software at all. They lean on experience. Texas McCombs’s Johnny Moreno commented, “Obviously, we’re going to be able to see and tell if you use AI.”
Carnegie Mellon Tepper’s McGrath made the most strategically useful version of this point — the interview is where an AI-scaffolded application collapses: “As you get to that interview stage, if you used an AI tool to help you with idea generation, coming up with your stories… the likelihood is the interview’s not going to go well, things aren’t going to make sense, your answers will be really shallow.”
The Parts of the Application that Are AI-proof by Design
Pay attention to a deliberate trend: Schools are building components AI simply cannot help you fake.
Video and spoken elements are the clearest example, and several schools said so openly. UVA Darden introduced a video essay this year: a randomly selected prompt from a bank of roughly 100, 10 seconds to review, 90 seconds to record, one take. Asked directly whether the tight timing was meant to defeat AI, West confirmed: “Exactly. It’s very intentional.”
Kellogg expanded to five video-essay prompts precisely because, as Katie Dyssegard noted, “you can’t even use AI as a tool to help you answer those.”
London Business School and Yale both drew a hard boundary at anything spoken. LBS’s Kirsten Wells noted, “One thing that we are very strict on is when we get to an interview process. That is where AI is absolutely not allowed.” Yale’s Botelho cautioned, “Stay away from it for any of the spoken elements of your application.”
Certain written prompts are engineered to be AI-resistant, too. Duke’s “25 Random Things About Yourself,” Cornell’s returning “Life Story” table-of-contents essay, Berkeley’s “what makes you feel alive” video — these are, as Graham put it, “kind of AI-proof,” because a generic model has nothing genuine to say about your specific life. Berkeley’s Millette remarked on their prompt, “With AI, it would be really hard to answer this question.”
One School is Turning AI into the Essay Itself
The most forward-looking development came from Georgetown McDonough, which is replacing one of its essays with an AI-focused prompt for the coming cycle. Crucially, it’s not a technical test. “It’s not going to be about the technical use of AI,” Kilby explained. “Instead, it’s going to be about the responsible use, the ethical use, the value creation… the discernment.” It reflects a broader direction several schools hinted at: AI fluency is becoming something they teach and assess, not just police. Darden and SMU Cox both noted they train students on AI tools once enrolled. The skill they want to see isn’t prompt-engineering — it’s judgment about when and how to use it.
What to Actually Do: A Practical Playbook
Pulling the five sessions together, here is how to use AI without hurting yourself.
Write the first draft yourself.
Every school’s line in the sand is content generation. Start from your own reflection — your stories, your reasons, your voice. LBS’s Wells said, “Try and make it not the first starting point.”
Use AI as a back-end editor, not a front-end writer.
Grammar, typos, tightening, flow. Columbia’s Murphy said, “We really want you to be your most human self here, and would encourage you to not use AI to generate the ideas for you. Take your most genuine approach to writing your essays yourself, and use AI to ensure they’re accurate, no typos, things like that.”
If a school asks you to disclose your AI use, disclose it.
Megan Lewis of Washington Foster noted, “Our policy is that you can use it however you’d like, but we want you to tell us how you used it.” Alessandra Gazzola said, “In terms of AI, we don’t have an official policy per se, but we do strongly recommend candidates disclose if they use AI.”
Run the reverse-prompt test.
A tip Graham raised in nearly every session, endorsed by Foster’s Lewis and UNC’s Hertz: once your essay is done, paste it into an AI tool without the prompt and ask what question it thinks you’re answering. If the answer doesn’t match the school’s actual question, you’ve drifted — a common failure when candidates recycle one school’s essay for another. This is a genuinely smart use of AI.
Have a real human read it.
The most repeated non-AI tip across all five webinars. Berkeley’s Millette said, have “a close friend, someone you know really well, go over your essays… they’re going to be able to tell you whether or not it really does sound like you.” Georgetown’s Kilby suggested handing a friend your essay without the question to see if they can guess it — the human version of the reverse-prompt test.
Protect your voice above everything.
If you remember one sentence, make it Tuck’s Harrison: “If a tool is creating your content, you sound like a computer. You don’t sound like you, and authenticity is really what’s going to get you admitted.”
Two AI Tools Built for Exactly This
Used correctly, AI can genuinely strengthen your candidacy — as a coach and a sounding board, not a ghostwriter. Two Clear Admit tools are purpose-built for the job:
Ask Clear Admit is an AI assistant trained solely on Clear Admit’s two decades of admissions knowledge — including the very webinar conversations quoted throughout this article — rather than the open internet. That focus matters: as Cornell’s Harlow warned, general-purpose models often “pull really old information,” citing programs and classes that no longer exist. Ask Clear Admit draws on current, admissions-specific material, making it a far more reliable place to research schools, pressure-test your goals, and sanity-check your understanding of what a program actually offers.
The Clear Admit Interview Simulator lets you practice under realistic conditions before you ever face a live interviewer. Given how many schools now use timed video responses and interviews specifically to surface the authentic, unscripted you — and given Tepper’s warning that AI-scaffolded essays fall apart at the interview stage — rehearsing your real answers out loud is one of the highest-leverage things you can do. This is AI helping you show up as yourself, which is exactly what every admissions officer said they’re looking for.
The Bottom Line
Twenty-three admissions officers, five webinars, one message: AI is a welcome assistant, but should not be an author. Use it to polish, to research, to rehearse, and to check that you’ve answered the question. Do not use it to think for you or write for you. The application is your one chance to sound like a person these readers want in the room — and, as Booth’s Chen reminded everyone, they are not admitting ChatGPT.
Quotes have been edited for length and clarity.
