Still in Charge · Insights

Everyone Thinks Everyone Else Is Cheating

By Thomas Tornatore, Founder · November 8, 2026

The student doing the work often believes they are the only one left. Some of the gap is students hiding what they do. A good part of it is a story bigger than the truth, and a parent or teacher can check it.

"Everyone's using it."

If you have a teenager, or teach a room full of them, you have probably heard some version of that sentence this fall. Sometimes it comes out as a defense. Sometimes it is a complaint about a classmate who got the better grade. And sometimes it comes out quietly, from a student who has been doing the work the slow way and is starting to wonder why.

On October 30, Dartmouth published a survey that put that sentence to the test. It asked 1,003 college students across the country how they use AI, and then how they think their classmates use it. Eleven percent said they often use it to get problem-set answers without understanding them. Forty-nine percent believe their classmates do. Eight percent said they often paste in AI writing with little editing. Half believe their classmates do.

Those first numbers are what students admit to, and people underreport what they are not proud of. Even so, the distance between what students say they do and what they believe everyone else does is wide. Your student is living inside it.

The honest kid thinks they are alone

Think about the student who is doing the work unassisted. Every night they choose the slower way, believing they are one of a few still doing it. That is a lonely place to be at sixteen. After a while it stops feeling like integrity and starts feeling like being the only one who did not get the memo.

Sooner or later, some of them give in. In the same survey, 39 percent of students agreed that they feel they have to use AI on assignments "because so many of my peers do so." When Dartmouth asked students about to arrive on campus to think back on the high schools they had just left, 14 percent said the same.

And the pressure does not end with school. Whether your student heads to college, a trade, the military or straight into a job, AI will be part of the work waiting for them. At many companies, using it is already commonplace. What will matter there is not whether they used it. It is whether they can still do the work themselves and tell when the tool got it wrong. That is built now, at your table, while the stakes are a homework grade.

The students are not confused about the cost. Eighty-four percent agreed that using AI for writing can hurt the development of their writing skills. They know. What wears them down is not a lack of information. It is the sense that everyone else has already decided.

Their frustration is fair

Do not tell your student it is all in their head, because it is not. Some classmates really are handing their homework to a machine, and they are being sold the practice as well. A post in my feed showed a young man in a suit, sitting at a desk with flags behind him, over the words "Claude finished my thesis in 78 hours and my advisor didn't even notice," and a promise of the prompt that did it. Posts like that make handing off the work look normal, easy and invisible. Your student sees them too.

When the researchers asked students to explain the gap, the answer was "an even mix of not wanting to self incriminate plus an overestimation of the peer cheating rate."

So your student's frustration is fair. Seventy percent of the students surveyed said take-home work has become unfair because so many students use AI on it. A kid who did the work and watched someone else coast has a real grievance, and it deserves to be heard before it is answered.

But even after the researchers allowed for students hiding the truth, the gap stayed large. Everyone is not everyone. The honest kid is not nearly as alone as they think.

What you can say tonight

Verify Outputs, our third principle, is usually about checking what a chatbot says. "Everyone is doing it" deserves the same check.

Ask before you tell. "What share of your class do you think uses AI to do the homework for them?" Let them answer without correcting it. Then tell them what college students said about themselves, and ask whether it surprises them.

Ask who. "Everyone" is impossible to argue with. Three names are not. "Who do you actually know is doing it? What did you see?" Often it comes down to one or two loud examples, and the rest is talk. The researchers wondered about this too: students may be picturing the worst thing they have seen, not what is typical.

Ask the real question. You do not need to convince your student that AI can stunt their writing. Most already agree. Ask instead, "If you knew most people were doing their own work, would it feel different?"

Say what you see. If your student is doing the work, tell them you noticed, and that it matters to you more than the grade. A kid who believes they are the last honest one in the room needs to hear that someone is watching for the right thing.

What you can do in your classroom

Your honest students may believe they are the minority in your room. You can find out. Ask two anonymous questions: how often do you use AI to do the work for you, and how often do you think your classmates do. Share the results with the class. Whatever the numbers show, your students will be looking at their own room instead of a rumor.

Then say out loud which work you value and why. The student who feels foolish for doing it the hard way needs to hear, from you, that the hard way is the point.

Not as alone as they think

The most useful thing you can give a teenager under this pressure is not another rule. It is the knowledge that they are not alone.

So when your student says everyone is using it, do not argue. Ask how they know.

What the checking turned up

Seven things about the evidence behind this piece, because you should be able to go look.

The social media post described above was labeled by the platform as AI content, and the man in it appears to be generated, not a real student. We have no way to check whether anyone wrote a thesis that way, and we do not name the account. It is described here as an example of what is being promoted, not as a true story.

This is a survey of college students, not high school students. The national sample of 1,003 was weighted and drawn through College Pulse from students at four-year colleges, and fielded in September 2026. A separate group of about 1,200 current Dartmouth undergraduates gave lower numbers on both sides of the gap: 9.5 percent said they use AI for problem-set answers, and 41.6 percent believe their peers do. Applying any of it to a high school student is this piece's argument, not the survey's finding.

The survey asked which tasks students "frequently" use AI for, and which tasks their peers frequently use it for. Occasional use is not counted on either side. In the Dartmouth version of the survey, published with the working paper, the box for peers comes before the box for the student's own use, and students are asked how often their peers use AI before they are asked about themselves. The national survey was "similar, though not identical," and its wording is not published.

Every figure is self-reported. To allow for students hiding the truth, the authors doubled every self-reported figure, based on earlier research suggesting people underreport sensitive behavior by about 40 percent at most, and found the gap still large. That is a rough adjustment, not a separate measurement. They also say they cannot rule out that students are describing the worst behavior they have seen rather than typical behavior.

The 14 percent comes from 324 incoming Dartmouth students answering before they arrived, in a survey the researchers call "similar, though not identical," with the wording tailored to the school each student had just left. They are a different group from the national sample, so the comparison shows a direction, not a change measured in the same students.

In one passage, the working paper describes the pressure finding as students feeling they "have to cheat to keep up with their classmates." The survey statement itself says "use AI," not "cheat." This piece quotes the statement.

The classroom survey suggested for teachers is this piece's idea. The Dartmouth survey did not test whether sharing the real numbers changes behavior.

Dartmouth AI Survey 2026, first results: ai.dartmouth.edu/dartmouth-ai-survey-2026

Sacerdote and Parris-Bacon, working paper, October 29, 2026: ai.dartmouth.edu/sites/ai/files/2026-10/AIs%20Impacts%20in%20Higher%20Education%20Sacerdote%2010.29.26.pdf

Dartmouth news release, October 30, 2026: home.dartmouth.edu/news/2026/10/dartmouth-survey-reveals-what-students-think-about-ai

When we get something wrong in a published issue, it will be listed here too.