UNAM, Mexico’s largest university, is moving a major admissions decision back into an exam room after its first fully remote entrance test produced results that looked sharply different from previous years.
Nearly 160,000 applicants took the entrance exam earlier this summer over several weeks from late May through early June. The test used a “lockdown” browser and AI-powered webcam proctoring software, but the outcome has now forced the university into a much larger correction: about 58,000 people could be affected by a new in-person control exam.
The score pattern triggered the crisis
The central issue is not simply that the exam was online. It is that the results changed dramatically, especially among applicants with the strongest scores.
Between 2021 and 2025, 3.5 percent of test takers scored 100 or more on the 120-question UNAM test. This year, 16.3 percent reached that level. At the very top, the difference was even sharper: between 2021 and 2025, 0.9 percent scored 110 or more, while this year 5.5 percent did so.
Those numbers created immediate questions about whether the remote testing system had protected the integrity of the admissions process. The surge in top scores led to accusations of widespread cheating, and UNAM appointed a commission of experts to review what had happened.
The commission was formally named “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 was direct: applicants should take a “control exam,” and that exam should happen in person.
Who has to retake the exam
The new test does not apply only to applicants who had already secured a place based on this year’s results. It also applies to everyone who would have been admitted based on minimum successful scores in their program of study since 2021.
That means about 58,000 people could be affected. Places at UNAM will now depend on the results of the new test, not simply on the disputed remote exam results.
The timing is tight. Details on the control exam were expected soon, while classes were scheduled to begin on August 10. Unless UNAM delays classes, the university has little room to resolve the problem slowly.
According to Gaceta UNAM, the school’s official news publication, the university rector apologized to honest applicants who now have to prepare again despite doing nothing wrong. The rector also said the control exam is “necessary to give certainty and guarantee equity in access.”
What the remote system was supposed to prevent
The remote exam was multiple choice, which made some familiar signs of AI-assisted cheating harder to identify. There were no essay answers where investigators could easily look for blocks of text pasted into answer boxes. The possible methods raised in reporting included AI assistance, cheat sheets, leaked questions, and other traditional approaches.
A New York Times report said 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.”
The security setup was intended to reduce exactly those kinds of risks. A report in El País said the test required the LockDown Browser from Respondus. According to the company, once an exam begins, students are unable to print, copy, move to other web addresses, use other computer applications, run web searches, use instant messaging, or minimize the browser. After the exam, the computer returns to its normal state.
UNAM also used a proctoring system from Territorium. The software monitored each applicant’s webcam using AI algorithms and looked for signs such as another person replacing the applicant, use of cell phones or earphones, or the applicant leaving the camera frame.
Human oversight was also part of the process. The test had one human supervisor for every 150 applicants. The system could send alerts to supervisors, who could notify UNAM officials about irregularities.
Why AI proctoring did not settle the issue
The measures appear to have caught some conduct problems. UNAM canceled nearly 2 percent of total exams for unspecified conduct issues. But that figure did not resolve the larger concern created by the score distribution.
UNAM has acknowledged “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.”
The case shows a practical problem for AI-supervised exams. A lockdown browser can restrict activity on the test computer, and webcam monitoring can flag visible behavior. But those tools cannot automatically prove that a high score is legitimate, especially when the overall results shift so far from prior years.
For applicants, the burden is immediate. Many people who followed the rules still have to take the entrance exam again because the process as a whole no longer gives UNAM enough confidence. For the university, the in-person control exam is now the mechanism for restoring certainty around who receives a place.
The final admissions outcome will depend on that new test. After an experiment in fully remote testing, nearly 58,000 applicants are being sent back to a traditional setting: an official exam room with a human proctor watching in person.