How I read 150 schools’ AI policies, and how to catch me being wrong

I check every school’s official, applicant-facing pages: dedicated AI policy pages where they exist, application instructions, admissions FAQs, admissions blog posts, and academic-integrity pages the school ties to admissions. I record the exact quote, the URL, and the date I read it. If a classification and a quote ever seem to disagree, the quote wins. Corrections: email me and I will fix it and log the fix at /changes.

What does not count as a source: news stories, other websites’ summaries, forum posts, or anything a school said in private. If it is not on the school’s own pages, it is not in the index.

Which schools

The 150 are the schools in Second Read’s prompt database: the selective national universities and liberal-arts colleges that ask for supplemental essays, the large public systems, and the schools international applicants ask about most. It is not a ranking and it is not every college. The list is the index itself.

When

The full read of all 150 schools was done August 27, 2026. I re-read every page again on September 8, 2026; a row whose page could not be reached that day keeps its earlier date, and every row shows its own date. Through the season I re-check pages when a school publishes something new or a reader reports a change, and any change to a quote, a classification, or a date is logged publicly. I do not re-read every policy page every week, and I do not claim to.

The five classifications

Effectively prohibits (7 of 150)

The school’s published text limits AI to basic proofreading or less, says AI may not be used in preparing application materials, or requires you to certify that no AI was used. “Effectively” is my word, and it is an inference: several of these schools allow a spell-check or bouncing an idea off a tool, and none of them names “feedback” as such. What they rule out, on my reading, is what Second Read does, which is feedback on a draft. These are the schools where Second Read refuses school-specific reviews; the product shows the school’s words instead. I quote the language rather than paraphrase it, so you can read it differently.

Published guidance (36 of 150)

The school has published rules for AI in the application: some uses allowed, others not. Many of these schools explicitly permit brainstorming, grammar help, or feedback on a draft you wrote, and forbid writing, rewriting, or translating. A few warn without clearly permitting anything. The line between this class and the one above is whether the restriction reaches feedback on a draft; where a page could be read either way, I read it the stricter way.

Commentary, no restriction (5 of 150)

The school has published thoughts about AI in the essay process without restricting how you use it. The quote is the whole of it, and it is not permission for anything in particular. The certification you sign when you submit still applies. 3 of these (The University of North Carolina at Chapel Hill, University of Massachusetts Amherst, Virginia Polytechnic Institute and State University) describe the school’s own use of AI in reading applications and say nothing about yours; the row says so.

No published policy (102 of 150)

I found nothing about AI on the school’s official admissions pages on the day I checked. This is its own class on purpose. An earlier draft of this index folded these schools in with the permissive ones, and that would have been misleading: silence is not permission. What binds you at these schools is the certification you sign when you submit, which says the work is your own, plus the school’s general integrity language. The row links to the page I checked so you can see the silence for yourself.

Not yet verified (0 of 150)

I have not located and read an AI policy page for these schools. I do not guess. Second Read pauses school-specific reviews there until I have, and the row says so. None at the moment: the three schools that started the season here (Fordham, whose page bounces automated fetches to a login gateway, and UMass Amherst and Virginia Tech, whose pages blocked us) were read in a browser on the re-read.

Whose words

For 7 schools the quoted text needs a caveat before you read it: it belongs to one part of the school rather than the whole (an honors college, a page for international applicants, a dean’s blog), or it describes the school’s own use of AI rather than a rule for applicants. Those rows carry a note saying so: Princeton University, Texas Christian University, The University of North Carolina at Chapel Hill, University of Massachusetts Amherst, University of Missouri, Virginia Polytechnic Institute and State University and Wesleyan University.

“But Kaplan says 68% of colleges have no AI policy.”

My read agrees with Kaplan. Kaplan’s November 2025 survey asked 220 admissions officers whether their institution has an official policy on AI in application essays; 68% said no. Going school by school through these 150, I found nothing published at 102 of them, 68%. Different method, same result. What this index adds is the other third: the exact words of the 48 schools that do say something, and which 7 of those I read as prohibitions. It is also why this site does not hand every applicant a “disclosure statement”: at most schools there is nowhere to file one, and volunteering a statement nobody asked for invites scrutiny with no upside.

Limits, stated plainly

These classifications are one person’s readings of public text on given dates. Policies move mid-season. Some schools ask about AI inside the application itself rather than on any public page, and an index built from public pages cannot see that; read your application’s own questions. That is why every row carries a verbatim quote, a source link, and the date I read it, and why the changelog exists. This is not advice, and I am not your counselor. Verify with the school.

Check the work

The whole index is a free CSV, college-ai-policy-index-2026-27.csv: one row per school with classification, verbatim quote, scope note where one applies, source URL, the date I read it, and whether Second Read refuses school-specific reviews there. Re-score all 150 yourself. If you land somewhere else on a school, write to support@getsecondread.com; admissions offices writing from a .edu address get a fix or an annotation within 48 hours, logged publicly.