We found no college policy that says admissions officers discount a project because AI helped build it. The published rules target something else: submitting AI-written application text as your own. So an AI-built project isn't a problem in itself, but the description you submit must be your own writing, and you should be able to explain every decision in the project if a reader or interviewer asks.
Will admissions officers discount a project built with AI?
You built a tool with an AI coding assistant. It works, people use it, and it may be the most interesting thing on your application. You're wondering whether a reader will see "built with AI" and mentally cross it out.
Most pages answering this question are about essays, and many come from companies selling admissions help. So we went to the sources that matter: the admissions policy pages at Caltech, Notre Dame, Brown and Yale, plus an Associated Press report on how colleges are using AI.
Here's the honest finding. None of them says how readers weigh a project built with AI. There's no published rule that it counts for less, and none that it counts the same. Anyone who tells you confidently either way is guessing. What the sources do show is where the line sits, and it isn't where most students assume.
What do college AI policies actually cover?
All four policies are about the material you submit, not the tools behind your outside work.
Caltech's guidelines for fall 2026 applicants are written about essays. Its reason is simple: "Your essays are where we hear your voice." Yale's policy focuses on written responses too, and has applicants affirm that what they submit is their "own work, factually true, and honestly presented."
They share a foundation. Yale, Notre Dame and Brown all point to the Common App's fraud policy, which Notre Dame quotes. It treats intentionally misrepresenting "the substantive content or output" of an AI platform as your own original work as fraud.
Read that wording closely. The problem it names is misrepresentation. Our reading, and it's only a reading, is that a project you openly describe as AI-assisted isn't misrepresenting anything. A project you present as entirely hand-built might be. If you're relying on that distinction at a particular school, ask its admissions office.
So there are two separate things to keep apart:
- The project: the tool you built, outside the application, with whatever help you used.
- The application text: the activity line, essay or short answer you write about it.
The policies speak clearly about the second. They're nearly silent on the first.
Where do the policies disagree?
They differ on how far the rule reaches, and on what help is allowed.
Scope. Caltech's guidelines name essays. Notre Dame goes wider, requiring that "essays, short answer questions, activities, and any other material" be your own work. Brown also covers any other submitted material. So at some schools your activity description is explicitly inside the rule.
Allowed help. Brown is the strictest of the four: AI use with application content is "not permitted under any circumstances," with basic spelling and grammar review as the exception. Notre Dame adds brainstorming before you write. Yale allows grammar and spelling review, plus general advice or topic suggestions at the start. Caltech allows grammar review of a finished essay and questions to get brainstorming going, but not AI outlines or drafts, and it also rules out using AI to translate an essay you wrote in another language.
The practical result is that the safest habit is also the simplest. Write every word about your project yourself, and treat a spelling check as the upper limit until you've read each school's page.
These pages are written mainly for undergraduate applicants, and they change by cycle. Graduate programs may have their own rules, so check each one.
How would a reader test whether the project is really yours?
The most useful clue comes from Caltech, and it isn't about tools at all.
The Associated Press reported in December 2025 that Caltech was launching an AI tool for students who submit research projects. Applicants upload their research, a chatbot interviews them about it on video, and Caltech faculty review the recording. The admissions director described the question behind it: "Can you claim this research intellectually?"
That concerns research rather than student-built apps, and it's one school. But the test carries over well. It doesn't ask who typed the code. It asks whether the work lives in your head. The same AP report says some selective schools are using AI to vet curated application packages built with consultants, which is the same concern from another direction.
This is our inference, not a published standard: a project will lose value in a reader's eyes when you can't claim it, and that's true whether a consultant, a parent or an AI did the part you can't explain.
Checklist: before an AI-built project goes on your application
- Read each school's current AI policy. Check which materials it covers and what it allows. Don't rely on a summary, including this one.
- Write the description yourself. Your words, your claims. Run a spell-check only where it's allowed.
- Name the tool in plain words. "Built with an AI coding assistant" is enough. Don't imply you wrote it all by hand.
- Lead with your decisions. What problem you picked, who you talked to, what you cut, what you fixed.
- Use only numbers you can show. Real users, real dates, screenshots or analytics kept somewhere safe.
- Practice a two-minute walk-through. How it works, what broke and how you checked it. If you can't explain a part, learn it before you submit.
- Keep the record. Commit history, notes and early versions show how the project grew.
- Treat required portfolios separately. Work a program asks you to submit is application material, so read its rule before using any AI-generated piece.
Compare two activity lines:
| Weak | Stronger |
|---|---|
| Founder of an AI-powered study platform | Built a quiz tool for my bio class with an AI coding assistant; 14 classmates used it before finals; rewrote scoring after it failed on partial answers |
The numbers there are only an example. Use yours, even if they're small.
How STARC handles this
STARC's one-person company cohort assumes you'll build with AI, and asks you to say how. Each weekly build log covers what shipped, the numbers, what users said, where you're stuck and how you used AI that week. Logs need evidence for their numbers. The cohort leaves you with a live product, real usage data, eight logs and a case study, and no certificate.
That doesn't promise anything about admissions, and we never do. The first cohort hasn't run yet, so there are no results to point to. The rules are on the How it works page, and the FAQ covers who can apply.
The short version
No published policy says an AI-built project counts for less. The rules cover what you submit, so write about the project in your own words and say plainly how you built it. Then make sure you can claim all of it when someone asks.
Related questions
Do I have to disclose that I used AI to build my project?
None of the policies we read requires it for a project. Say it anyway, in plain words, unless a school gives other instructions. A project that sounds hand-coded and turns out not to be is a much bigger problem than one line about your tools.
Can I use AI to write my activity description?
Don't. Notre Dame says activities must be your own work, and Brown extends its rule to any material you submit. A grammar check is allowed at those schools, but the words and claims should be yours.
Is a project built with AI still impressive to colleges?
That depends on what it shows. Nobody has published how readers weigh this, but a working tool with real users and a clear account of your decisions says more about you than which software wrote the code.
What if a school's AI policy doesn't mention projects?
Most don't. Read the school's current policy, follow its rule on written materials, and email the admissions office if you're unsure. Policies change from one cycle to the next.
Want to build your own product?
The OPC cohort: 12 students, 8 weeks, one real product with real users.
Follow STARC on X. New posts, what we're building, and the questions students send us: @Starc_Institute
Follow on XLast updated: 2026-11-03. Written by the STARC team. Rules quoted here come from our how it works page; if they change, this post is updated.
More questions
What is a build in public cohort, and is it right for students?
What a build in public cohort actually is, what students gain and risk by joining one, and the questions to ask before you sign up.
What can I build with AI in 48 hours? One-page tool ideas for beginners
What a beginner can really build with AI in 48 hours: one-page tool ideas, a weekend schedule, and what to leave out so you actually ship.
Should I put vibe coding on my resume as a student?
Should you put vibe coding on your resume? Why the phrase is risky, what to write instead, and how to show AI-assisted work honestly.