UNAM orders retest after AI-cheating scandal in admissions exam

UNAM halted undergraduate admissions procedures and activated a 16-member technical commission after unusually strong scores on its first-ever online entrance exam raised fears of widespread cheating. The National Autonomous University of Mexico, the country’s largest and most prestigious public university, was weighing whether to rerun the admissions test as scrutiny intensified.

The trigger was the exam’s score pattern. Officials and reports pointed to results that looked too strong to be ordinary, with one analysis suggesting nearly half of test-takers may have cheated. Another account put the number of potentially involved students at about 75,000, a figure that captured the scale of the breach and the strain it placed on the admissions system.

UNAM rector Leonardo Lomelí Vanegas said the university’s priority was protecting applicants who took the exam honestly and clarifying the irregularities. He also said the university would seek measures to guarantee equitable access to higher education, signaling that administrators saw the problem as more than a single testing failure and as a test of admissions fairness itself.

The scandal centered on UNAM’s first online undergraduate admissions exam in July 2026, a shift that immediately exposed the university to the same vulnerabilities now affecting schools around the world: remote testing, easily shared answers, and AI tools that can help students evade traditional controls. Coverage around the case pointed to chatbots and writing tools such as ChatGPT and Grammarly as part of the broader challenge, where the line between assistance and cheating has become harder for institutions to police.

Public pressure built quickly. Rejected applicants protested outside the UNAM Rectorate Tower in Mexico City on July 28, pressing the university to account for how the exam was run and whether the results could still be trusted. By July 30 and 31, UNAM was publicly considering whether to validate the online scores or rerun the admissions exams.

For Mexico’s most influential public university, the episode went beyond one admissions cycle. It raised a harder question for universities that rely on high-stakes entrance exams: if AI can be used at scale, and if authorities can no longer trust the numbers that determine who gets in, the old model of testing may no longer be defensible without stronger safeguards.