The Wharton School MBA application AI policy, 2026-27

AI may not substantially write your essay

Wharton's rule is an analogy with teeth: treat AI the way you would treat another person. You could not have a friend substantially write your essay, so you cannot have a model do it either. Wharton pairs the rule with a stated willingness to run AI-detection tools, and a flag triggers a fuller human review of your whole application. Bounded help is permitted; substantial authorship is not.

What Wharton's policy says, in our reading

This is our plain-English reading of Wharton's published materials, not the school's own words. Read the official source before you rely on it.

The load-bearing phrase is substantially write. Feedback, structural suggestions, and the kind of critique you would accept from a colleague sit inside the policy. Drafting does not. The person-analogy is genuinely useful as a self-test: if you would be uncomfortable telling Wharton a friend did this exact thing for you, do not let a tool do it.

Wharton is the only school of the twelve that says out loud it may run detection software. Detection tools are imperfect, and Wharton seems to know that: a flag buys scrutiny, not rejection. The practical consequence is that flattened, generic prose is a double liability here, since it can trip a detector and reads badly to humans even when it does not.

Wharton states its position inside the live application rather than on a public page; confirm there. · Sourced Jul 14, 2026

No disclosure form exists, so the checklist is the person-test

Run the substitution test on every AI interaction: would this have been acceptable coming from a human helper?
Keep the drafting loop closed: your words go out for critique, and only your words come back in.
De-flatten before you submit: idiosyncratic detail and specific numbers are what tired readers and generic-text detectors both notice missing.
Log your tool use privately; if HBS or Ross is also on your list, the same log feeds their disclosures.

Wherever a school asks for a disclosure, the statement itself should be yours. Our free compliance check audits how you used AI against Wharton's rules and shows what your own statement needs to cover; it never writes it for you.

Run the free AI-policy compliance check →

How Wharton compares to the other 11 schools

Wharton sits between the poles: firmer than Kellogg's and Columbia's bounded-use policies, softer than Stanford's flat prohibition.
It is the only school that states it may run AI detection, which makes voice authenticity weigh more heavily here than anywhere else.
Like Tuck, Wharton frames its rule through existing norms (another person, an honor standard) rather than a new AI rulebook.

The full field, side by side, lives on the MBA AI-policy hub. Wharton's current essay prompts are on the prompt tracker.

Frequently asked questions

How much AI help counts as substantial at Wharton?

Wharton does not quantify it; the person-analogy is the guide it offers instead. A reviewer pointing at weak paragraphs is help; a rewritten paragraph is authorship. When a specific case feels arguable, assume it lands on the wrong side.

Does Wharton actually run AI detectors on essays?

Wharton says it may, and publishes no detail on which tools or how often. Treat it as real but secondary: the humans reading your file are the detector that matters, and they read hundreds of essays a week.

What happens if my Wharton essay gets flagged?

A flag leads to a closer review of your application rather than automatic denial. That is exactly why the risk is asymmetric: an authentic essay survives scrutiny, while a machine-assisted one invites questions that are hard to answer well.

Schools with a similar stance

Stanford GSB AI policy guide · Prohibits AI-written essays
Tuck AI policy guide · Essays must be exclusively yours
CBS AI policy guide · Ideas and editing yes, complete responses no

School policies change between and within cycles. This page reflects Wharton's public materials as of the sourced date above; always confirm against the live application before you rely on it.

Loading…