Mexico City — More than 58,000 students at the National Autonomous University of Mexico (UNAM) are being forced to retake a remote exam after the online proctoring system failed. The incident has thrown the reliability of AI-based invigilation into doubt and sparked a broader debate about digital surveillance in higher education.
What went wrong with the remote exam
The university had deployed an AI proctoring tool to monitor students during a high-stakes online test. But the system malfunctioned, flagging thousands of test-takers for suspicious behavior that later turned out to be false positives. UNAM administrators determined the results were too compromised to stand. They ordered a complete retake for all 58,000 students who sat the exam.
Students reported being locked out of the test interface, receiving error messages, and having their cameras or microphones cut off mid-exam. The AI software, designed to detect cheating by tracking eye movements, background noise, and unusual activity, instead flagged routine actions like looking away to think or pausing to type. The scale of the failure left the university with no choice but to scrap the original scores.
AI proctoring under the microscope
The UNAM disaster is the latest case to raise questions about the effectiveness and limitations of automated proctoring. Critics have long warned that these systems are prone to bias, especially against students with disabilities, those in poor lighting, or those who fidget. The incident at UNAM shows that even when the technology works as designed, it can still produce unreliable results under real-world conditions.
Privacy advocates also point to the invasive nature of AI proctoring. Students are required to share their screen, allow camera and microphone access, and often submit to room scans. The UNAM case adds fuel to arguments that such surveillance may not be worth the trade-off in student privacy and trust.
Reevaluating digital surveillance in assessments
The event has prompted a reevaluation of digital surveillance in academic assessments at UNAM and beyond. University officials are now reviewing their contract with the proctoring vendor and considering alternative approaches. Some faculty members have called for a return to in-person exams or the use of open-book, project-based assessments that don't rely on constant monitoring.
Other institutions across Latin America are watching closely. Many adopted remote proctoring during the pandemic and have kept it in place. The UNAM failure could accelerate a shift away from high-stakes online testing or at least push for stricter oversight of the technology.
For now, the 58,000 students must prepare to sit the exam again. The university has not yet announced a new date or said whether it will use the same proctoring system. The question hanging over the campus: can AI proctoring ever be trusted at scale?
