Overview
- A viral case from Alcorn State University showed 32 of 35 students failed part of a midterm after professor Jason Gibson hid a white‑text instruction that forced AI to insert nonsense about Madagascar, exposing mass prompt‑copying and inattentive proofreading.
- A second viral example from a schoolteacher using a hidden 'Gatorade' prompt produced similar results when an AI checker flagged essays that had copied the full prompt into a chatbot.
- The incidents, widely reported starting Monday, have renewed scrutiny of AI‑detection tools after a HEPI report found many of those apps are inaccurate and some universities have stopped using them.
- Universities are shifting policy by restricting devices in class, returning to supervised or handwritten exams, and promoting staged, process‑based assessment and oral defenses to make student authorship visible.
- Students and educators face new fairness and access questions because traps can fail (for example when students use dark mode) and detectors can produce false positives for non‑native writers, so appeals and clearer AI literacy rules are becoming central to campus responses.