TL;DR: The Caltech AI policy treats AI the way a lab treats outside help: assistance is legitimate, authorship is not transferable. Its admissions guidance permits using AI the way you might use a person, for researching the school, brainstorming directions, and checking spelling and grammar, while ruling out having the tool outline, draft, write, or translate your essays. Around that guideline sits a verification layer: Caltech has incorporated a short video or personal explanation component, giving readers a direct look at the applicant behind the file. The combination is coherent and very Caltech: define acceptable collaboration precisely, then build an instrument that checks the signal. Backing it all is the certification you sign, under which submitting generated language as your own is misrepresentation regardless of any named rule.
Sources: Caltech admissions guidance; Common App fraud policy.
Help Like a Human Could Give: Reading the Guideline
Caltech’s test is portability: if the help would be legitimate coming from a teacher, counselor, or friend, it is legitimate coming from a tool, and if it would cross the line from a person, it crosses the line from a machine. A friend can point you to Caltech’s core curriculum quirks, react to your topic ideas, and catch your typos; a friend cannot ethically outline your essay, draft your paragraphs, or translate a piece you wrote in another language into the submitted English, because at that point the friend is the author. Caltech’s guidance maps AI onto exactly that framework, which makes it one of the most teachable policies in selective admissions: the question is never what the tool can do, it is whose writing lands on the page. In practice, the Caltech AI policy mirrors its human help test: what a trusted mentor could legitimately do, a tool can do, and nothing more.
The Ledger: Sanctioned Assistance and Transferred Authorship
| Use | Caltech’s Position | The Human Help Test |
|---|---|---|
| Researching Caltech programs, houses, and core | Sanctioned | Information gathering a person could provide |
| Brainstorming essay directions | Sanctioned | Reacting to your ideas, not replacing them |
| Spelling and grammar review | Sanctioned | Mechanics on writing that is already yours |
| Outlining the essay | Ruled out | Structure is authorship at Caltech |
| Drafting or writing any portion | Ruled out | The author must be the applicant |
| Translating your draft into submitted English | Ruled out | The submitted language must be yours |
Notice how close this sits to Dartmouth’s think partner framework, with one Caltech signature addition on the enforcement side, covered next. The full eighteen school map lives in our college AI policies overview.
The Verification Layer, and the STEM Applicant’s Trap
Caltech pairs its guideline with an instrument: a short video or personal explanation component that lets readers see and hear the applicant directly, an authenticity check no text detector can match. For Caltech’s applicant pool the design carries a particular warning. Strong STEM students are the demographic most fluent with these tools and most tempted to treat essays as a formality to be automated, but Caltech’s writing prompts exist precisely because the institute wants evidence of a mind, not a specification sheet, and a generated essay beside an unscripted video of the real person is the easiest mismatch in admissions to spot. The winning approach is the honest one: write like the person who does the problem sets, use the tools only where a friend could help, and let the video sound like the essays. Our guide to AI detection and acceptable use covers the broader evidence landscape.
Frequently Asked Questions About Caltech and AI
Caltech allows AI only as human style help: researching the school, brainstorming directions, and checking spelling and grammar are sanctioned, while outlining, drafting, writing, or translating your essays with AI is ruled out. The test is whether the same help would be legitimate coming from a person.
No. Translation is explicitly on the ruled out side of Caltech’s guidance, because the submitted English must be your own writing, not a rendering of it. Multilingual applicants should draft in English themselves and use sanctioned help only for mechanics.
A short recorded or written verification layer that gives readers direct, unscripted access to the applicant, functioning as an authenticity instrument beside the essays. Treat it as a voice match: the person on camera should sound like the person on the page.
Because the numbers at Caltech’s level stop differentiating: nearly every serious applicant presents maxed rigor and scores, so the writing and the verification layer carry the evidence of curiosity, collaboration, and motive. Automating the one differentiating surface is the most self defeating shortcut available.
The certification you sign treats submitted work as your own; generated language submitted as yours is misrepresentation, with consequences from denial to withdrawal of an offer. Caltech’s honor culture makes the authorship standard continuous from application through enrollment.
Sources: Caltech Admissions, Common App, NCES College Navigator, IPEDS, NACAC
About Oriel Admissions
Oriel Admissions is a Princeton-based college admissions consulting firm advising families nationwide on elite university admissions strategy. Our strength is a deeply experienced team and a distinctive 360 approach that treats every part of the application – academics, testing, activities, essays, and interviews – as one connected strategy. To discuss your strategy, schedule a consultation.
