Most advice about AI and college applications is written by people guessing. The universities themselves have published rules, and the rules disagree with each other so sharply that the same action can be encouraged at one university and grounds for revoking your admission at another.
So we read them. Below is what six US universities and the Common App publish, quoted from their own pages, with the link so you can check each one yourself. We have put the update date on anything that carries one, because these pages are being rewritten every year. The UK has taken a different approach, and that is at the end.
First, the thing to understand before any of it: most universities have said nothing at all. Kaplan surveyed more than 200 admissions officers between July and August 2025 and found that 2% of colleges have a policy allowing applicants to use generative AI to write their essays, 30% ban it, and 68% have no official policy either way. For brainstorming, 27% allow it, 4% ban it, and 69% have no policy. Silence is the most common position, and silence is not permission.
Common App: AI output is named in the fraud policy
This one binds you at over a thousand universities at once, whatever any individual college says.
The Common App fraud policy lists, among the circumstances in which fraud may occur:
Submitting 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 platform, technology, or algorithm
Note the word "substantive", which the policy does not define. Note also what the policy does not do: it does not carve out an exception for grammar tools, and it does not tell you where assistance becomes substance. The consequence is set out in the same document, which says Common App "reserves the right to limit, restrict, or terminate access to the Solution by any Registered User alleged to have engaged in Fraud or other misuses on a temporary or permanent basis."
Losing your Common App account in December would end your application cycle.
Brown: nothing beyond a spellchecker
Brown is the strictest of the universities we read. Its statement on integrity in the application process says "the use of artificial intelligence by an applicant is not permitted under any circumstances in conjunction with application content."
The only thing it allows is using AI "to assist with spelling and grammar review, in the same way as any other platform that supports basic proofreading", and it requires that "the content of all essays, short-answer questions and any other material submitted by an applicant must be the work of that individual."
Applicants certify that what they submit "is my own work, factually true, and honestly presented", and acknowledge they "may be subject to a range of possible disciplinary actions, including admission revocation, expulsion, or revocation of course credit, grades, and degree" if that certification is false.
Caltech: do not translate your own essay
Caltech publishes a list of acceptable and unacceptable uses, and the unacceptable list contains a line that should stop every international applicant:
Translating an essay written in another language
Alongside it: "Copying and pasting directly from an AI generator", "Relying on AI generated content to outline or draft an essay", and "Replacing your unique voice and tone with AI generated content".
What Caltech does allow is narrower than most students assume but wider than Brown: "Using AI tools, like Grammarly or Microsoft Editor, to review grammar and spelling", "Generating questions or exercises to help kick start the brainstorming process", and "Using AI to research the college application process".
The test it offers is a good one to carry around: "Would it be ethical to have a trusted adult perform the same task you are asking of ChatGPT?" Failure to comply "may result in the rescission of your admission to Caltech".
Northwestern: the most specific list published
Northwestern's guidance, updated 23 September 2025, is the most specific list published by any university we read.
Acceptable: "Researching colleges and admission terms", "Building a college tour itinerary", "Exploring how your academic interests and goals connect to fields of study", "Brainstorming essay topics", and "Reviewing the grammar, punctuation, and spelling of your completed essays".
Not acceptable: "Using generative AI to write your personal statement or supplemental essays", "Providing generative AI with an essay outline and submitting AI's completed piece", "Using generative AI to rework a supplemental essay that was written for one school for another school", and, again, "Translating an essay written in another language".
Two universities on opposite sides of the United States ban the same thing in the same words, and it is the thing a multilingual student is most likely to do.
Notre Dame: revocation or expulsion
Notre Dame's AI policy permits AI "to assist with spelling and grammar review, or brainstorming support before writing".
It prohibits "submitting content generated, drafted, or substantially written by artificial intelligence tools or other text-generating software", and states that doing so "may result in the revocation of your offer of admission or potential expulsion".
Georgia Tech: the most permissive American position
Georgia Tech went first, publishing guidance in July 2023 written by its admissions director with a colleague from the College of Computing, while most institutions said nothing.
It is the most permissive American position we found. Applicants are told to "use it to brainstorm, edit, and refine your ideas", that AI "can also be a useful tool as you consider how to construct your resume", and that tools can help "generate ideas, provide alternative phrasing options, and organize your thoughts".
The line it draws is at the copy and paste: "you should not copy and paste content you did not create directly into your application".
Read that against Brown, where any use beyond proofreading is "not permitted under any circumstances". A student applying to both is being given opposite instructions about the same draft.
The UK does it differently
Everything above is American. If you are applying to the UK as well, which most of our students are, the rules change shape entirely, because the UK runs on a single platform that has published its own position.
UCAS's guidance is more permissive than any US university on this page. It tells applicants they must declare "that your personal statement hasn't been copied or provided from another source, including artificial intelligence software". Then it tells them how to use AI anyway. UCAS suggests using it to generate ideas "which you can then relate to your own experiences", to "suggest ways of structuring your personal statement answers", and to "check your personal statement draft for readability".
Asking AI to structure your statement is advice from UCAS and a breach at Brown.
What UCAS rules out is "generating (and then copying, pasting and submitting) all or a large part of your personal statement from an AI tool such as ChatGPT, and presenting it as your own words", which "could be considered cheating by universities and colleges".
The enforcement is where the UK diverges completely. Every statement goes through similarity detection, and the applicant guide explains what that means: statements are checked against "a library of personal statements previously submitted to UCAS", samples collected from websites, and other sources including paper publications. "Any statements showing a level of similarity of 10% or more are reviewed" by UCAS staff, and applicants and universities are notified at the same time when similarity is confirmed. Your statement joins the library after it is processed, so it becomes part of what next year's applicants are checked against.
Ten percent is a lower bar than most people assume, and AI tools produce exactly the kind of generic phrasing that thousands of other applicants are also being handed.
Cambridge: a made-up achievement is fraud
Cambridge's page allows AI "as prompts for ideas or to assist in research when preparing your personal statement", and then raises a risk nobody else does.
AI "may generate false claims ('hallucination')", and Cambridge is explicit about the consequence: "If these are about your personal achievements, they will be treated like any other application fraud."
A student who asks a chatbot to improve a paragraph and does not check what it added back can end up claiming a prize they never won. The university will not treat that as a software error.
Cambridge also addresses the interview, where "any use of AI tools during an interview hinders this approach and is therefore prohibited", and says interviewers use "various methods" to "randomly check that AI tools are not being used".
Imperial: the video interview is being watched
Imperial Business School's admissions guidance permits "getting tips on how to organise your personal statement", "catching typos or awkward phrasing", and "generating prompts to reflect on your experience or career goals", while insisting that "the ideas, the reflection and the tone? That should be your own".
For the recorded video interview it is blunt: "Reading from prepared answers or relying on AI tools during the recording is not permitted." It adds that "we've introduced new ways to detect AI-generated content and are continuously adapting our review process", and that reading "is immediately visible and creates a stilted delivery".
What this means if English is not your first language
This is the part of the story that is written for somebody else and lands hardest on our students.
Two of the most specific policies published, Caltech's and Northwestern's, ban "translating an essay written in another language" in identical words. Not translating with AI badly. Translating at all, as a listed unacceptable use.
For a student who thinks in Uzbek, Arabic, Portuguese or Hindi, drafting in your first language and then carrying it into English is the obvious way to get your own voice onto the page. Two universities have decided that the result is not your work. Meanwhile UCAS, reading the same question from a country where a third of applicants write in a second language, is recommending AI for readability checks.
There is no single rule to follow here, so the practical answer is to write in English from the start for any university that bans translation, and to use the tools for the things every policy on this page permits: understanding what a university wants, generating questions to think about, and checking spelling at the end.
The students whose essays worked did not have a translation problem to solve, because the material was theirs. One student wrote about translating documents at the Embassy of Pakistan and then building a program to do it, and that became the Princeton personal statement. A student from Turkey wrote about cooking for the family through a mother's illness. A student from Egypt wrote about a love of building things, and another wrote about a childhood injury that led to hearing loss and then to computers. A Barnard student wrote about her grandmother. One wrote about enrolling in a theatre club without telling their parents, and won a full scholarship to Northwestern with it.
No tool can supply any of that.
What to do before you submit
Check each university separately. The policies are not converging. Brown, Caltech, Northwestern, Notre Dame and Georgia Tech all publish different lines, and 68% of colleges publish nothing.
Treat silence as the strict rule, not the loose one. If a university has published nothing, the Common App fraud policy still applies to your application, and it names AI output directly.
Never let a tool add a fact. Cambridge will treat an invented achievement as fraud whether or not you noticed it being added.
Read everything aloud before you submit. If a sentence is not one you would say, an admissions reader who has read ten thousand essays will notice, and so will the interviewer who meets you later.
We build an AI counselor, so we should say where we think the line is. A tool is useful for the parts of this process that are research and logistics: what a university costs, which deadlines are open, whether your grades fit, what the policy on this page says. The essay is the one part of the application that exists to find out who you are, and handing it over defeats the purpose of writing it.
Start your own journey
Open the admissions page for every university on your list and search it for "artificial intelligence" before you write a word. It takes a few minutes and it is the only version that counts, because these pages change between cycles.
Kai, the Borderless AI college counselor, helps with the research, the deadlines and the shortlist, and will tell you what a university expects rather than writing your essay for you. It's free to use.
Start with Kai, and read the stories from students who wrote their own.
