On Campus, More AI Use Means More Cheating. Across Majors, It Means Less

A Science analysis of 95,513 students at 20 public research universities estimates that about 9% of AI users submitted AI-generated work they knew might violate course rules.

Generative AI use is much more common than cheating, and adoption differs sharply by major while cheating rates stay relatively narrow across disciplines.

Frequent AI users were more likely to cheat than occasional users, suggesting misconduct rises with heavier personal use.

The findings offer one of the first large-scale estimates of AI-related cheating in higher education and show that AI adoption and misconduct do not move in lockstep.