US College Professor Embeds Hidden Instruction in Midterm, 32 of 35 Students Submit AI-Marked Answers

A US college history professor embedded a hidden instruction in a midterm discussion-post assignment, telling AI tools to include the word "Madagascar" nonsensically in responses. According to USA Today, Professor Jason Gibson said 32 of his 35 students submitted answers containing the hidden word, and those students failed a portion of the midterm. Gibson said he does not oppose students using AI, but draws the line at having AI generate an entire assignment, and believes schools and educators need to adapt to AI rather than ban it.

The episode turns AI-use detection from abstract debate into a verifiable classroom operation: a teacher can embed a hidden instruction and observe whether students paste an entire prompt into AI and submit the output. For teachers and schools, it signals the need to define where AI use is allowed and where it is not, and to design assignments that expose unread submissions; for students, understanding and reviewing submitted work remains a basic requirement. The finding comes from one professor's classroom practice as reported by media, with a sample of only 35 students, and cannot be generalized into a universal cheating rate or a learning-outcome conclusion. Teachers gain a low-cost classroom method to flag wholesale AI-generated submissions, though hidden instructions may also affect students who use AI as a legitimate aid. Schools and course designers need clearer boundaries between AI assistance and AI ghostwriting, along with requirements for reviewing submitted work.