The State of AI in College Admissions (gradpilot.com)

A new survey maps how 166 U.S. universities treat generative AI in college applications, categorizing institutional guidance into five clear buckets: No Explicit Policy, Permissive/Integrative (AI allowed with accuracy responsibility and often disclosure), Line-level Help (grammar/style edits like Grammarly okay but not wholesale drafting), Brainstorm Only (outlines and idea generation permitted; rewriting banned), and Prohibited (no AI use at all). Policies were collected directly from official admissions sites, portals, and communications, with source URLs and verification dates included; the dataset is updated around admissions cycles and users can request additions or corrections via support@gradpilot.com. This matters because applicants, counselors, and admissions officers now face a patchwork of rules that affect essay authorship, disclosure expectations, and fairness assessments. Technically, the taxonomy highlights where institutions expect human final wording versus where machine-aided composition is acceptable, which has downstream implications for detection tool use, integrity enforcement, and equitable access to AI-assisted drafting tools. For researchers and admissions teams, the dataset offers a verifiable baseline to study policy trends, harmonize guidance, and design workflows or detection methods aligned with each policy type.