AI Glasses in Exams: Why Banning Devices Alone Won’t Stop Cheating—and What Parents and Teachers Can Try
The notification popped while I was wiping down the stove after dinner. South Korean exam proctors were reportedly seeing students use AI-equipped glasses to photograph test questions, feed them into artificial intelligence tools, and receive answers through tiny earpieces or hidden displays. I stood there for a second, sponge in hand, looking at my son hunched over his laptop at the kitchen table.
“Did you know people were using glasses to cheat on exams?” I asked.
He didn’t look up. “Yeah. A kid in my stats group said smart glasses are going to be impossible to catch.”
That answer bothered me more than the news itself. Not because I think every teenager with a device is cheating, but because the response was so matter-of-fact. The technology had already arrived in his imagination. The adults, meanwhile, were still arguing about where phones should be kept during a quiz.
The South Korean cases are easy to treat as a distant, high-tech scandal. But the problem they expose is ordinary. Most schools still rely on the same basic exam model: students sit in a room, devices are banned or placed in baskets, and everyone answers the same questions under time pressure. That model assumes the only internet access a student has is a phone in a pocket. Smart glasses, earbuds, watches, and other wearables make that assumption look fragile.
One morning last month, I proctored a history midterm. We collected phones in labeled envelopes. We checked desks. We walked the rows. A student wore glasses connected to a small armband case, and I had to quietly ask whether they were regular prescription lenses. They were. The student was embarrassed, and I felt awful. But the moment showed the trap schools are now in. If we start treating every pair of glasses as suspicious, we invade privacy and target students who need assistive technology. If we ignore wearables, some students will use them to cheat.
Policing the devices is necessary in a limited way. During a closed assessment, my rule is now plain: no phones, smartwatches, earbuds, or camera-equipped glasses on anyone’s body. Prescription glasses are fine; students who use assistive technology work with the learning support team ahead of time so accommodations are arranged privately. The phone basket stays by the door. But those are guardrails, not a solution.
The real issue is that many tests still reward exactly what AI does best: recall, formula application, short factual answers, and predictable essays. If a grade depends mostly on repeating information from a study guide, a tool that can retrieve and organize that information will always be tempting. A student who is exhausted, afraid of losing a scholarship, or convinced that “one bad grade ruins everything” may not pause to think about the ethics.
I learned this two years ago with a student named Daniel. He was quiet, usually prepared, and taking three advanced courses. During a vocabulary quiz, I saw him glance at his lap, where his phone was tucked under his hoodie. I took the phone, gave him a zero on the quiz, and called home. Later, when we spoke, he didn’t make excuses. He said he had fallen behind after a soccer tournament, needed an A, and saw ten seconds on his phone as the only way through.
That didn’t make the cheating acceptable. But it made me see that my quiz design had made the shortcut obvious. The words came from a fixed list, the questions were similar to the review sheet, and the quiz was a large part of his grade. I could have spent the semester searching hoodies and pencil cases. Or I could have changed what the quiz asked students to do.
After that, I started breaking assessments into smaller pieces. Some quizzes are open-note and ask students to apply a term to a new situation, not just define it. For larger tests, part of the grade comes from a short explanation at my desk. I might pull two or three index cards with student names and ask, “Walk me through why you chose this answer,” or “Show me where this idea appears in your notes.” It takes only a minute per student, but it changes the calculation. Memorizing or receiving an answer is less useful when you may have to explain your thinking.
I also change the context of questions each term. Instead of asking students to label the causes of an event from the textbook, I give them a short new document and ask which cause it best supports. That is harder to outsource because the answer depends on material seen for the first time in the room. It isn’t perfect, and it takes more planning, but it is more honest about what we should be assessing anyway.
At home, the same principle applies. Parents do not need to become AI experts or install surveillance software on every device. Small routines matter more. When my son studies, I sometimes ask him to flip his phone face-down on the counter and teach me one question without looking anything up. If he cannot explain it in his own words, he does not understand it well enough yet. I also ask what the assessment actually requires. Is he being asked to memorize dates, compare sources, solve a new problem, or write a personal response? Those conversations help him see when AI is a useful practice tool and when using it would simply hide what he hasn’t learned.
Schools can support this by separating two questions that often get mixed together: “Did you use AI?” and “What did you learn?” Some assignments should explicitly allow AI as a tutor or editor, with students required to submit the prompts they used and explain how the response changed their thinking. Other assessments must be device-free, but they should be designed so a device cannot do the essential work. A blanket ban without that distinction teaches students mostly how to hide tools.
When cheating does happen, the response should be serious but specific. A zero, a conversation, and a chance to redo a different assessment can be more useful than a permanent label. Ask what made the risk feel worth taking: Was the test too high-stakes? Was the student behind? Did they understand the rule? That does not excuse dishonesty, but it helps adults see the pressure underneath.
The South Korean glasses cases will not be the last story of their kind. Devices will get smaller, faster, and more ordinary. Schools can keep updating banned-item lists, and some of that is unavoidable. But chasing every new gadget is exhausting and unfair, especially when assistive technology and ordinary medical devices become tangled in suspicion.
Last week, my students completed an open-note document analysis. Phones were in the basket, and their notes were spread across their desks. Several students struggled, but I could hear them arguing about what the source meant. That was messy, slow, and harder to grade than a multiple-choice test. It also felt closer to the thing I actually want to assess.
I still lock up phones during exams. I still watch for earbuds and unusual devices. But I no longer pretend that enforcement alone solves the problem. The glasses are new. The pressure to find the fastest answer is not. Until tests value thinking, explanation, and transfer more than quick recall, students will keep looking for an easier route—and technology will keep giving them one.
Please indicate: Thinking In Educating » AI Glasses in Exams: Why Banning Devices Alone Won’t Stop Cheating—and What Parents and Teachers Can Try