The line most schools draw
Policies vary by school and are still being written, so the only safe rule is to read the AI policy on each application you file. That said, the published ones converge on one distinction: assistance versus authorship. Feedback, critique, proofreading, and structural suggestions sit on the same side as a writing centre tutor or a mentor reading your draft. Generating prose you then submit as your own does not — most applications ask you to certify the work is yours.
This is the same standard applied to human help for decades. A friend who tells you your second paragraph is vague is fine. A friend who writes your second paragraph is not.
Why ghostwritten statements fail anyway
Beyond the ethics, generated statements read as generated: even syntax, no odd specifics, an emotional register that never varies. Readers who see hundreds of files a week detect it as flatness even when they cannot name it — and it removes exactly the thing the statement exists to show.
Safe uses of AI on your statement
- Asking what a committee reader would find vague, generic, or unsupported.
- Identifying which sentences could belong to any applicant.
- Checking structure: does the argument survive a skim?
- Grammar, punctuation, and consistency passes.
- Pressure-testing whether your fit paragraph names anything specific.
Uses that cross the line
- Asking for a draft, an opening paragraph, or a closing to paste in.
- Having a model invent experiences, numbers, or reflections.
- Rewriting your sentences into a voice you do not write in.
How Gradline is built around that line
Gradline highlights and comments. It marks the phrases a committee reader would cut or question, gives a blunt verdict, and lists concrete fixes — then leaves the revising to you. It does not produce replacement prose, and your statement is not used to train models or shown to anyone else.