UNAM's 2026 Online Entrance-Exam Fiasco — A Cautionary Case in AI-Assisted Cheating at Scale

In 2026, Mexico's National Autonomous University (UNAM) administered its undergraduate entrance exam entirely online for the first time in its history, testing roughly 158,000 applicants between May and June using a proctored web browser and AI-based webcam monitoring.
A commission of external experts reviewing the results found the score distribution statistically implausible: only 3.5% of test-takers scored 100 or more out of 120 questions between 2021 and 2025, but 16.3% did so in 2026; perfect or near-perfect scores rose from 0.9% to 5.5% over the same comparison.
Investigators concluded that widespread use of generative AI assistants — including ChatGPT, Gemini and Claude — was the most plausible explanation, and TikTok videos circulating beforehand offered a 'cheating manual' advising applicants to run a chatbot on a second monitor outside the webcam's frame.
UNAM ordered roughly 58,000 of the flagged applicants to sit a supervised, in-person 'control exam' rather than accept the original online results, reverting to pen-and-paper testing for the retake. The episode is now cited widely in higher-education circles as a cautionary case on the limits of AI-based remote proctoring at national scale.

Read the full analysis: https://www.npr.org/2026/08/15/nx-s1-5929156/the-ai-cheating-disaster-that-forced-mexicos-top-university-back-to-pen-and-paper