AI-supervised exam failure forces 58,000 students to retake test

A remote exam using AI supervision went wrong badly enough that 58,000 students must retake it, with top scores jumping fivefold during the flawed session. The report does not specify the exact cause of the AI monitoring or scoring failure.

Why it matters: Large-scale AI-proctoring failures like this feed growing skepticism about deploying AI oversight tools in high-stakes settings without robust human fallback, and could slow adoption of AI exam supervision elsewhere. It's a concrete data point in the broader debate over trusting AI systems with decisions they aren't yet reliable enough to make alone.