Between 2021 and 2025, 0.9 percent of applicants scored 110 or higher on UNAM‘s 120-question entrance exam. This year, 5.5 percent did.
That’s a six-fold jump at the very top of the distribution, in a single year, in the first year the test was given remotely. Nobody at Mexico‘s largest university needed a statistician to know something had gone wrong.
Now roughly 58,000 people have to sit the thing again, in a room, with a human watching.
The numbers that broke the exam
Nearly 160,000 applicants took the UNAM entrance exam earlier this summer, over several weeks stretching from late May into early June. It was the first time the test ran entirely remotely, using a lockdown browser and AI webcam proctoring.
The score distribution came back looking nothing like the previous four years. Between 2021 and 2025, 3.5 percent of test takers scored 100 or more. This year, 16.3 percent cleared that bar.
Scores don’t move like that. Not across 160,000 people, not in 12 months, not without something changing about the conditions.
What the software was supposed to stop
The test required Respondus’ LockDown Browser. According to the company, once an exam begins in this browser, students are “unable to print, copy, go to other web addresses, or access other applications on the computer. Web searches, instant messaging, minimizing the browser, and hundreds of other functions are also prevented. Once the exam is completed, students are permitted to exit the browser and their computer will return to its normal state.”
That’s a real lockdown, and it’s the standard product in this category. It also only controls one machine.
UNAM layered on a proctoring system from Territorium that ran AI algorithms against each test taker’s webcam feed. It watched for a substitute sitting down in front of the computer, for cell phones or earphones, for the applicant leaving the frame entirely. One human supervisor covered every 150 applicants, receiving system alerts and escalating irregularities to UNAM officials.
The workarounds were public before the test started
A New York Times report last week noted that “cheating tips were circulating widely before the tests went live this spring. Online accounts advised students to position monitors outside the web-camera frame where they could access ChatGPT or other AI models. Others suggested hiding earphones under their hair, or hiring someone else to take the exam, off camera.”
Read those three tactics against the two products above and the gap is obvious. A lockdown browser secures the computer running the exam. It has nothing to say about the second monitor two feet to the left. Webcam AI looks for earphones, so you cover the earphones with hair.
The countermeasures were designed for a candidate alone with one machine. The candidates weren’t.
How much of it got caught
UNAM canceled nearly 2 percent of total exams over unspecified conduct issues, so the system flagged some people. Set that against what the university itself concedes: “the probability that a significant number of applicants may have received help” on the test.
AI expert Raul Rojas, speaking to NPR, said statistical modeling suggested “that almost half of the students were cheating when taking the exam online.”
Two percent caught against a modeled 50 percent. Whatever the exact figure, that’s not a detection system with a tuning problem. That’s a detection system watching the wrong thing.
And nobody actually knows yet what happened. The exam was multiple choice rather than essay-based, which strips away the usual AI-cheating tells, like a complete answer appearing in a text box all at once. Cheat sheets, leaked questions and the rest of the analog playbook are all still on the table.
Who pays for it
UNAM appointed a commission of experts to investigate, saddled with the name “la Comisión Técnica de Personas Expertas para la Revisión del Proceso de Selección de Ingreso a Licenciatura para el Ciclo Escolar 2026-2027/1.” Its recommendation has landed: a control exam, in person.
The retake doesn’t stop at this year’s cohort. It applies to everyone who would have been admitted based on minimum successful scores in their program of study since 2021. About 58,000 people are in scope, and their places at UNAM now hinge on the new result.
Per Gaceta UNAM, the school’s official news publication, the rector has apologized to honest applicants, who will now prepare for and sit a test again having done nothing wrong. The control exam is still “necessary to give certainty and guarantee equity in access,” the rector added.
The clock is the other problem
Details on the control exam should appear soon. Classes are currently scheduled to begin on August 10, which leaves almost no room to schedule, staff and run an in-person exam for 58,000 people unless the school pushes the start date.
Here’s the part any institution shopping for remote proctoring should sit with. UNAM bought two products, added a human supervisor for every 150 candidates, and still ended up unable to certify a single score. The fallback wasn’t a partial audit or a re-scoring. It was a room, a proctor and 58,000 people doing it over.
