Princeton Survey Finds 29.9% of Seniors Cheated, Mostly With Generative AI
Updated
Updated · Ars Technica · May 13
Princeton Survey Finds 29.9% of Seniors Cheated, Mostly With Generative AI
6 articles · Updated · Ars Technica · May 13
29.9% of Princeton seniors in a 2025 survey said they cheated on at least one assignment or exam, with most of that misconduct involving generative AI.
40.8% of bachelor of science in engineering students admitted cheating, versus 26.4% of bachelor of arts students, pointing to heavier pressure in some programs.
Princeton’s 1893 honor code bars professors from proctoring in-class exams, leaving students to sign an honor pledge and report peers themselves.
Cell phones, AI tools and a campus culture resistant to “snitching” have strained that system, allowing cheating to spread while often going unreported.
In the fight against AI cheating, is Princeton sacrificing its historic culture of student trust?
If policing students fails, how must universities evolve to teach skills that AI cannot replicate?
Princeton Reinstates Proctored Exams After 133 Years, Citing AI-Driven Cheating Crisis
Overview
Princeton University will end its 133-year tradition of unproctored, honor-based exams and reintroduce proctored examinations starting July 1, 2026. This major policy change follows an overwhelming faculty vote in May 2026, driven by a 2025 senior survey that revealed high rates of academic misconduct, including cheating and unauthorized use of artificial intelligence. The survey also showed widespread non-reporting of Honor Code violations, highlighting a breakdown in peer enforcement. These findings underscore growing challenges to academic integrity at Princeton, prompting the university to combine proctoring with its traditional Honor Code to better uphold fairness and honesty.