What happened with AI-supervised remote exams?
A remote exam supervised by AI went awry, resulting in 58,000 students needing to retake it due to concerns over widespread cheating. Top scores increased by 5x, prompting an investigation and a decision to administer a control exam in person.
So, earlier this summer, Mexico's largest university, UNAM, administered an AI-supervised remote exam to nearly 160,000 applicants - and the results were pretty alarming. I mean, the exam itself was taken completely remotely, using a "lockdown" browser and AI-powered webcam proctoring software, which sounds like it should be foolproof, right? But somehow, top scores increased significantly compared to previous years - we're talking a jump from 3.5 percent of test takers scoring 100 or more between 2021 and 2025, to 16.3 percent achieving this score this year.
The surge in top scores led to accusations of widespread cheating, which prompted UNAM to appoint a commission of experts to investigate - and their recommendation is to administer a "control exam" in person. Now, this isn't just for those who secured a spot at UNAM based on this year's test, but also for everyone who would have been admitted based on minimum successful scores in their program of study since 2021. That's around 58,000 people who'll have to prepare for and take the test again, which is a pretty big deal.
The university rector has apologized to honest applicants, acknowledging that they'll have to take the test again despite doing nothing wrong - it's a tough spot to be in. The control exam is necessary to ensure equity in access and provide certainty, but it's still unclear exactly how the cheating happened. Maybe students relied on traditional methods like cheat sheets or leaked questions, or maybe they used AI models like ChatGPT to gain an unfair advantage - either way, it's a problem that needs to be addressed.
Source: Ars Technica
Related posts
New York Times Sues Brave Browser Over AI Scraping Paywalls
The New York Times and a group of publishers sued Brave Software on November 2, 2026, alleging the browser maker systematically scraped copyrighted articles without consent and sold access to AI companies including Anthropic, Perplexity, Mistral, xAI, and Cohere. The complaint also targets Brave's "Paywall blocker" feature, which grants access to hundreds of publishers without subscriptions. Publishers say Brave ignored repeated demands to stop and engineered crawlers to evade technica
Google Satellite Runs First AI Workload In Orbit
Google's Project Suncatcher satellite successfully activated its onboard Tensor Processing Units and ran the Gemma 4 model in orbit, generating a haiku about space. This marks the first time Google's custom AI silicon has functioned in the vacuum of orbit, proving the hardware survives launch, initializes in microgravity, and performs inference. The test validates space-based computing as a potential solution for Earth's data center constraints. Google plans a 2027 twin-satellite missi
Apple Brings Intelligence Features to China With iOS 27.3
Apple will launch Apple Intelligence in China via iOS 27.3, with beta arriving November 4 and public release in December. Regulators approved the suite in July. Features include notification summaries, translation, writing tools, visual intelligence, and photo object removal, but the redesigned Siri is excluded. Apple partnered with Alibaba to integrate Qwen chatbot and developed a China-specific large language model to comply with local regulations.