Every nightclub has that one bouncer checking IDs at the front door while somebody quietly props open the fire exit around back. UNAM, Mexico's most prestigious university, just found its own version, except the bouncer was an AI surveillance system built specifically to make its entrance exam cheat-proof, and the fire exit stayed open the whole time, propped by roughly 150,000 people's worth of foot traffic.

UNAM runs its entrance exam with an AI monitoring system watching in real time specifically to catch cheating. This year the results came back with an unusually high number of perfect scores, the kind of statistical anomaly that makes a testing office about as suspicious as a casino watching one table win forty hands in a row. An investigation turned up a genuinely retro toolkit of fraud sitting comfortably under the AI's nose: phones, companions, and identity theft, none of it exactly cutting-edge tradecraft. The bouncer was watching the front door so intently it apparently never looked down to notice someone walking in wearing a badge that wasn't theirs.

Then the fallout escalated well past the roughly 3,000 tests initially invalidated. Rather than only re-testing suspected cheaters, UNAM decided the only way to trust the results again was to make everyone start over, no AI, just pen, paper, and a proctor with actual eyeballs. That in-person makeup exam applies to anyone who passed well enough to enroll this year or in any of the past five years, close to 58,000 people, the overwhelming majority of whom did nothing wrong and are now retaking a test because the system built to vouch for them stopped being trustworthy, presumably including a fair number of people who haven't sharpened an actual pencil since the last time a Blockbuster was open. It's the academic version of the whole class losing recess because nobody will confess to the paper airplane, except recess is five years of college admissions.

UNAM's rector has since called for "ethical behavior" from applicants going forward, which is a wonderfully polite way of describing a scenario where the AI security system became the entire reason security had to be rebuilt from scratch without it. Presumably the same ethical behavior the AI was supposed to be enforcing on everyone's behalf the first time around.

The university built a bouncer smart enough to watch fifty thousand doors at once. It just never checked whether the bouncer could actually see the one that mattered, and now everyone's showing up for a second round of ID checks, this time with an actual human at the door who, unlike the AI, is legally required to blink.

Sources: CNN — The AI Cheating Scandal Rocking Mexico's Most Prestigious University (https://www.cnn.com/2026/08/02/americas/mexican-university-unam-ai-cheating-scandal-intl-latam); NPR via Public Radio East — The AI Cheating Disaster That Forced Mexico's Top University Back to Pen and Paper (https://www.publicradioeast.org/2026-08-15/the-ai-cheating-disaster-that-forced-mexicos-top-university-back-to-pen-and-paper)