Nearly 160,000 applicants sat for UNAM’s entrance exam in late May and early June 2026, the first time Mexico’s largest university administered the test fully online with AI-powered webcam proctoring. When results came in, the share of top scores had jumped by roughly fivefold compared with prior years. The university has now decided that about 58,000 applicants must retake the test in person, throwing the start of the school year into question.<\/p>\n
The 120-question UNAM exam is multiple choice, and its score distribution had been stable for at least five years. Between 2021 and 2025, 3.5 percent of test takers scored 100 or above, and only 0.9 percent scored 110 or above. In 2026, 16.3 percent reached 100 or above, and 5.5 percent reached 110 or above. The cluster of perfect and near-perfect scores was so far outside historical norms that the university concluded the test’s integrity could not be trusted.<\/p>\n
The size of the shift drew immediate attention. A 5x increase in top scores, and roughly a 6x increase at the very top of the scale, is not the kind of change that can be explained by a stronger incoming cohort. Something in the testing environment changed, and the candidates who took the exam remotely were the only group exposed to that change.<\/p>\n
UNAM appointed a commission of experts, formally named la Comisi\u00f3n T\u00e9cnica de Personas Expertas para la Revisi\u00f3n del Proceso de Selecci\u00f3n de Ingreso a Licenciatura para el Ciclo Escolar 2026-2027\/1, to review what went wrong. The commission’s recommendation was to hold a new in-person “control exam.” Applicants who secured a spot at UNAM based on this year’s results must take it, as must everyone who would have been admitted under minimum successful scores in their program of study going back to 2021. That pool adds up to about 58,000 people.<\/p>\n
UNAM classes are scheduled to begin on August 10, so the retake will have to be organized quickly or the calendar will slip. According to Gaceta UNAM, the university’s official news publication, the rector has apologized to applicants who did nothing wrong and now face a second round of preparation. The rector described the control exam as necessary to give certainty and guarantee equity in access.<\/p>\n
Because the 2026 exam was delivered remotely, UNAM required candidates to install testing-integrity software on their computers. According to a report in El Pa\u00eds, the test used Respondus LockDown Browser, which the company describes as preventing printing, copying, navigation to other web addresses, access to other applications, web searches, instant messaging, and browser minimization during an exam. Once the test ends, the computer returns to its normal state.<\/p>\n
On top of the lockdown browser, UNAM deployed a proctoring system from Territorium that monitored each candidate’s webcam with AI algorithms. The system looked for signs that someone other than the registered applicant had taken the seat, that a phone or earphone was in use, or that the candidate had stepped out of frame. Human supervisors were also assigned at a ratio of about one supervisor per 150 candidates and could receive system alerts about irregularities. UNAM canceled nearly 2 percent of exams for unspecified conduct issues, suggesting some cheating was caught, but the score data suggests much more went undetected.<\/p>\n
UNAM has acknowledged the probability that a significant number of applicants may have received help. The university has not identified a specific method, and because the exam was multiple choice, the usual signals of AI cheating seen in essay-based tests, such as answers pasted rapidly into text boxes, do not apply. Cheat sheets, leaked questions, and other traditional methods remain possible explanations.<\/p>\n
AI expert Raul Rojas, speaking to NPR, said that statistical modeling of the results suggested almost half of the students were cheating when taking the exam online. Reporting also described cheating tips circulating before the tests went live: advice to position monitors outside the webcam frame so ChatGPT or other AI models could be consulted, suggestions to hide earphones under hair, and offers to hire a proxy to sit for the exam off camera. The combination of a remote setting, a browser lockdown that is not bulletproof, and an AI proctor with known blind spots appears to have been too easy to defeat at scale.<\/p>\n
Every applicant affected will need to prepare for and sit an in-person test under human supervision. Spots at UNAM for the 2026-2027 cycle will now depend on the results of that control exam rather than on the contaminated June results. The university has not yet published logistics for the retake, and the August 10 start date leaves only days for the new testing operation to be assembled. A delay to the academic calendar is now a real possibility.<\/p>\n
For honest applicants, the cost of the failure is concrete: extra study time, more time away from work or family, and the stress of a second high-stakes test. For UNAM, the cost is a credibility hit to a flagship admissions process, and a reminder that moving a high-stakes, high-volume exam online introduces risks that current proctoring stacks cannot reliably handle.<\/p>\n
Top scores on the 2026 entrance exam jumped to 16.3 percent scoring 100 or above and 5.5 percent scoring 110 or above, compared with 3.5 percent and 0.9 percent respectively between 2021 and 2025. UNAM concluded the results could not be trusted and required a new in-person test.<\/p>\n
UNAM used Respondus LockDown Browser and an AI webcam-proctoring system from Territorium, with about one human supervisor per 150 candidates.<\/p>\n
Investigators have not identified a single method. Possibilities include leaked questions, cheat sheets, proxies taking the test off camera, hidden earphones, monitors placed outside the webcam frame for AI assistant access, and gaps in the AI proctoring itself. Statistical modeling cited in coverage estimated that almost half of remote test takers may have cheated.<\/p>\n