In 2026, Mexico's National Autonomous University (UNAM) faced a major crisis after implementing AI-powered remote proctoring for its entrance exam. Over 160,000 applicants took the test remotely using a lockdown browser and AI webcam monitoring between late May and early June. Results showed an abnormal spike in high scores, with 16.3% achieving 100+ on a 120-question test (up from 3.5% previously) and 5.5% scoring 110+ (compared to 0.9% before). This led to accusations of widespread cheating and prompted UNAM to establish a 100-member expert commission to investigate. The commission recommended administering a new in-person 'control exam' to all affected applicants, potentially impacting 58,000 students. The rector apologized to honest candidates, emphasizing the need for equitable access. Critics argue the AI system failed to detect cheating, with some suggesting vulnerabilities in the lockdown browser allowed circumvention. The incident highlights challenges in using AI for academic integrity monitoring, as the university now faces pressure to balance technological efficiency with fair assessment practices.