Professor Jason Gibson of the University of Mississippi embedded an invisible AI-detection prompt in a midterm exam to identify students using generative AI [1].
This method highlights the growing struggle for educators to maintain academic integrity as students increasingly rely on large language models to complete assignments. The use of "trap" prompts suggests that traditional plagiarism detection software may no longer be sufficient for modern classrooms.
Gibson, a history professor, placed a hidden command within the exam text that was invisible to the human eye but readable by AI tools [1]. When students copied and pasted the exam questions into a generative AI program, the AI followed the hidden instructions and included specific phrases or markers in its output [2].
These markers allowed the professor to instantly identify which students had outsourced their work to an algorithm. According to reports, 32 out of 35 students who took the midterm were caught using AI to generate their answers [1], [3].
The incident took place on the University of Mississippi campus in Oxford, Mississippi [4]. Gibson designed the prompt specifically to protect the rigor of his history course amid the rising accessibility of AI tools [1].
While some educators argue that such traps are necessary for fairness, others suggest that the prevalence of AI use indicates a need to redesign how students are tested. The high percentage of students caught in this specific instance, where only three students did not trigger the trap [1], underscores the scale of the challenge facing higher education in the U.S.
“32 out of 35 students were caught using AI to generate their answers”
The use of 'canary' or invisible prompts represents a shift toward adversarial testing in academia. As AI becomes more integrated into student workflows, professors are moving from passive detection to active entrapment to verify authenticity. This suggests a widening gap between traditional assessment methods and the capabilities of generative technology, potentially forcing universities to return to pen-and-paper exams to ensure academic honesty.
