Every change to the directory's data since it was created on July 18, 2026, newest first, with the schools it touched. A date is the day the data changed. A stance that moved is listed as such; a re-check that found nothing new is a full re-check. How entries are verified is on the methodology page.
· Schools added
Seventeen more high-volume schools added after their admissions pages were read on 2026-11-03, the second pass of the day. Chapman University publishes an admissions-office statement on ChatGPT and AI tools and enters as a verified permits-feedback policy. The other sixteen publish no AI rule for the general application and enter as checked schools with the pages read, the applications they accept and their own words on the essay and on honesty; two of them, Boise State and UNC Charlotte, bar AI for their Honors College essays, quoted on their pages. Each also gets a 2026-27 prompts entry: twelve listed or no-supplement, three Common App-only (unreadable), and Oklahoma State and Appalachian State read in a browser because their sites refuse automated readers. American University's site refused every reader a second time.
Nineteen high-volume schools added after their admissions pages were read on 2026-11-03. Two publish an AI rule and enter as verified permits-feedback policies: William & Mary (a dedicated AI and the Application Process page) and Colorado State (its admissions essay-tips post). The other seventeen publish none and enter as checked schools with the pages read, the applications they accept and their own words on the essay and on honesty. Each also gets a 2026-27 prompts entry: eleven listed or no-supplement, seven Common App-only (unreadable) and Colby read in a browser. American University's site refused every reader and waits for the next pass.
Six batch-11 entries re-read and re-dated 2026-10-25 (NC State short answer added, LSU and Nebraska honors prompts, Pitt sourced); notes tightened elsewhere without moving dates, so the oldest prompt check remains 2026-10-24.
Seven schools whose sites block automated reading were read in a browser; nine Common App-only schools recorded as unreadable and Tulane as having no current supplement, completing coverage of all 117 directory schools.
October sweep: all 117 entries re-checked and dated 2026-10-16; notes updated for Duke (its AI essay prompt is gone), UNC (FAQ now says the office does not use AI), Missouri, UNLV, Alaska Fairbanks, Princeton.
UC received a label saying its statement names AI only for readability help; BYU's label reworded for the new commitment; Swarthmore's label now lists translation among prohibited uses.
BYU's quote replaced after its page became an Essays and Activity Commitment (broader wording, stance unchanged); Swarthmore's quote expanded to include translation; ellipsis and punctuation fixes at five other schools.
Wesleyan added as verified permits-feedback from its international-applicants authenticity statement; SMU added to no-published-policy because its AI rules cover only a scholarship essay and graduate applicants.
Georgetown moved from advisory to verified as the fifth prohibits-all school once its application form became readable without logging in; Rice, USC, and Haverford moved from no-published-policy to verified permits-feedback.
Claremont McKenna added as the fourth prohibits-all school by owner decision: its application instructions limit AI to basic proofreading, the same structure as Brown's rule.
Coverage reached all 50 states: BYU added as the third prohibits-all school and UConn as permits-feedback; 44 public flagships and selective colleges added to the no-published-policy tier (16 carry checked_on 2026-08-11).
Expansion sweep added ten verified entries, including Villanova as the second prohibits-all school (writing supplements must be unaided by AI), and 27 schools to the no-published-policy tier.
Full re-sweep for the 2026-27 cycle: every verified, advisory, and no-published-policy entry (30 in all) re-read against its source and dated 2026-08-04.
Penn and Notre Dame moved from no-published-policy to verified (permits feedback): both published dedicated first-party AI pages after the July sweep, verified in a real browser.
Quotes re-transcribed verbatim from the live pages for seven schools; Michigan's quote and source URL replaced because it published new explicit guidance for 2026-27. No stance changed.
Per-school display labels added so each label claims only what the quote supports: Brown, Harvard, Columbia, Michigan, Washington University; the same day Caltech received one and Michigan's was rewritten for its new guidance.
Advisory tier created for policies known only from secondary sources: Georgetown's reported login-gated AI attestation and Princeton's dean's-blog caution; advisories are shown as cautions and never stop a review.
Top-30 sweep added 16 verified permits-feedback entries (verified tier now 18) and created a no-published-policy tier for 10 schools whose sites carry no applicant AI rule.
Directory created with two entries quoted from the schools' own published pages: Brown University (prohibits all AI use, including feedback) and Caltech (permits feedback, prohibits AI-drafted text).