A Hidden Trick Just Busted Almost a Whole Class for Using AI on Their Midterm

A Hidden Trick Just Busted Almost a Whole Class for Using AI on Their Midterm

A college professor found a clever way to catch students cheating with AI, and the story has spread fast online. It's simple, a little sneaky, and honestly kind of funny once you hear how it worked — and it says less about one clever trap than about how little some students bother to check the work they're handing in.

The Setup

Dr. Jason Gibson teaches history at Alcorn State University. He gave his students a midterm question about the Industrial Revolution — nothing unusual there. But hidden inside the question, written in white text that blends into the background, he added a secret instruction. It told any AI reading the question to randomly bring up Madagascar in a way that made no sense at all. Students couldn't see the hidden text with their own eyes. But if they copied the question into a chatbot, the AI would read every word, including the invisible part, and follow it.

What Happened Next

Out of 35 students across two classes, 32 got caught. They copied the question, pasted it into a chatbot, then pasted whatever came out straight into their answer box without reading it. That's how essays about factory workers and steam engines ended up with random Madagascar lines mixed in — one answer wandered into the strange claim that Madagascar floats sideways through the afternoon, and another blended a discussion of inequality with a line about a purple bicycle whispering to the ceiling. If even one student had skimmed their own answer first, they probably would have noticed something was very wrong.

The Professor Went Viral

Gibson posted about the whole thing on TikTok, and it took off. He seemed less angry than stunned that so many students trusted an AI's output enough to submit it without reading it once. He said he wasn't trying to embarrass anyone, and after confronting the students, he gave all of them a chance to challenge their grade if they felt it was unfair. Only two took him up on it. Everyone caught in the trap failed that part of the exam.

Why This Story Is Hitting a Nerve

This isn't really a story about Madagascar, or even about one professor's clever trick. It's touching a bigger nerve: teachers everywhere are struggling to figure out who's actually doing their own work anymore. It isn't an isolated case either. Earlier this month, a professor at Brown University let students take an exam at home and watched average scores jump from the usual 65 to 80 percent range up to 96 percent. He guessed that nearly every student in his class of 86 had used AI to cheat, except for two.

Conclusion

Gibson's trick didn't rely on fancy detection software. It relied on a simple truth: if you don't read your own work, you have no idea what it says. That gap is the real story here — not that students are using AI, but that some have stopped checking the results entirely. As AI tools get easier to use, more of these stories are likely to surface, and more professors will keep inventing tricks like this one to find out who's still doing their own thinking.