A History of Cheating in College: From Honor Codes to AI

By Adrian Gonzalez · Reviewed by Matthew Brewer, Ph.D., CEO of PracticeQuest

A timeline of college cheating methods

Cheating in college isn't something new. What changes is the methods students use to cheat, whether that is AI, answer sites like Chegg, or sharing test answers with other students through text chats. Over time, educational institutions have tried to address cheating through a variety of different systems. Each system was built to close the cheating problem in front of it, but each failed to account for the future methods of cheating students would develop.

The cheating crisis has only gotten worse with AI. Students use AI not just to complete homework, but to cheat on exams. In spring 2026, Brown University professor Roberto Serrano's economics course drew 86 students, roughly triple its usual enrollment, after he announced a take-home exam format. The take-home midterm averaged 96 out of 100, far above the course's historical range of 65 to 80, and Serrano suspected that most of the class had used ChatGPT (Whitford, 2026b). He voided the midterm and moved the final in-person. Eighteen students dropped the course, nine more never showed up for the final, and the average fell to 48.6, the lowest of his career (Pascual, 2026; Whitford, 2026b). AI is the most recent and most powerful method of cheating, but it is far from the first, and the history of the systems built to stop the earlier methods helps explain why this one is proving so hard to contain.

Honor Codes: Asking Students to Police Themselves

The oldest answer American colleges came up with was to stop watching students altogether. That sounds like an obviously bad idea today, but it made a kind of sense in context. In the 1800s, when the honor code system first emerged, going to university was largely reserved for the wealthy elite. During this period, upper-class men often held a strong sense of honor that universities used to enforce honor codes and prevent cheating. Cheating was heavily frowned upon; those who did cheat and were caught would be ostracized by their peers and have their social lives ruined (Glanzer & Cockle, 2021). Of course, this system was hardly perfect and failed on many occasions.

The U.S. Military Academy at West Point built one of the strictest versions of this idea. Cadets pledge that they will not lie, cheat, steal, or tolerate those who do, and a cadet-run honor system investigates any violation of that pledge. West Point is the United States's most prestigious military school, and breaking the honor code was no minor offense. Still, the system failed spectacularly in 1976. An instructor noticed identical answers on a hard electrical engineering assignment and traced it back to widespread, organized collaboration on work that was supposed to be done alone. About 150 upperclassmen resigned or were expelled, which remains the largest cheating scandal in the history of any U.S. service academy. A commission led by former astronaut Frank Borman investigated the matter on behalf of the Secretary of the Army and ultimately recommended that the expelled cadets be allowed to return (Secretary of Army Will Act on Cadets, 1976). Borman's commission found that the honor system had not been applied fairly. Expulsion was the only available punishment, which left the student-led committees confused about what should even count as cheating when the penalty was that severe. The commission concluded that investigating and punishing cheating was better left to school administrators.

Forty-four years later, in the spring of 2020, West Point's next major cheating scandal hit a STEM course again, and it happened online: 73 cadets, almost all of them freshmen, were accused of cheating on a remote calculus exam given during the pandemic, after instructors caught matching irregularities in the answers submitted (Watson, 2021). The pledge cadets recite didn't change between 1976 and 2020. The only thing that changed was the exam format.

An honor code is a statement of values and a pledge to uphold them. It can deter students who already share those values, but both West Point scandals show its limits: the code holds until the exam format makes cheating easy, and reporting a classmate carries a social cost that no pledge removes.

AI has now pushed even the most storied honor system past that limit. In May 2026, Princeton's faculty voted to place proctors in every in-person exam, ending a 133-year-old tradition of unproctored, student-policed testing, because AI had made cheating too hard to detect without supervision in the room (Whitford, 2026a). The school that trusted its honor code longer than almost any other decided that watching the room was necessary after all.

Proctoring: Watching the Room, then Watching the Screen

In the 1850s, Harvard began using standardized examination booklets for written finals. This began a shift away from oral exams toward a system with a different set of controls: a fixed, countable set of pages a student couldn't smuggle material into or out of. Scantron came later, for multiple-choice testing; it used optical mark recognition to make grading bubble sheets fast enough for large-lecture testing to be practical (Bogardus Cortez, 2016).

Both of those technologies solved a logistics problem, not an integrity problem. Exam booklets and Scantrons were built with the expectation that an instructor or proctor would be standing in the room watching students take the exam. When COVID pushed classes online, new problems arose. In 2021, Dartmouth's Geisel School of Medicine accused 17 students of accessing course material on Canvas during remote exams, based on activity logs pulled from the learning platform. The school ended up dropping every charge and apologizing, after an outside technical review found that the same logs could be produced by a device quietly syncing course material in the background while a student wasn't even using it (Alonso & Akbarzai, 2021). Schools learned the hard way that automated cheating-detection tools are often unreliable and can produce false accusations against innocent students.

Answer Sharing: From the Filing Cabinet to the Search Bar

Long before the internet existed, fraternities and sororities kept test banks: filing cabinets full of old exams, collected and passed down by members. Reusing an old exam was common enough that it gave students in the right social network a real advantage. In my own recent experience as a student, most students don't run cheating networks anywhere near that sophisticated, but people often told friends in later sections the answers to questions they had already seen.

In 2007, things changed when University of Texas at San Antonio freshman Alex Baldwin took the fraternity test bank system and put it on the internet. He scanned his own fraternity's test files and posted them to a password-protected website that grew into Fratfolder.com, open to any fraternity willing to contribute its own archive (Smith, 2012). The site initially drew scrutiny over copyright rather than academic integrity, since the exams belonged to the professors who wrote them. But it laid the foundation for every answer-sharing service that followed. Chegg and similar sites were created to provide detailed explanations for homework and test problems. In practice, many students don't use the platform to understand anything. They use it to find the answers, skip the explanations, and copy down whatever earns the best grade.

One of the larger Chegg cases hit the University of British Columbia's Okanagan campus, where chemistry professors found that students had consulted Chegg solutions during online final exams in several chemistry courses (Alden & Ha, 2020). Chegg didn't create a new way to cheat; it used the internet to sell access to the oldest one. The fraternity test bank took years to build, and access was limited to members. An answer site could turn one student's upload into an answer key for any future student willing to pay.

Group Chats: The Exam Becomes a Group Project

Messaging apps removed the delay. In 2017, Ohio State investigated 83 students in a business course over a GroupMe thread used to share graded assignment answers (Bever, 2017). The same pattern showed up anywhere group chats already existed for legitimate coursework coordination, including in STEM courses at Georgia State University, where the campus paper reported a case involving a GroupMe used to circulate quiz answers in an organic chemistry class. The professor responded by threatening the entire section with a zero on the affected quiz (Jones, 2020).

That organic chemistry case shows the same mechanism that later reappeared with AI, at a smaller scale. Group texts turned cheating from an individual decision into a shared, low-effort system. Students no longer had to seek out a service or take any real risk; they just had to join a chat. And once cheating became a group activity, it came with social pressure attached: even a student who never sought out answers could feel they were falling behind the classmates who had them.

What Doesn't Change

Every system in this story solved the cheating method in front of it and left the next one untouched. None of them addressed why students cheat. For most students, a degree now stands between them and a livable income, and every graded assignment reads as a step toward it or away from it. Under those stakes, cheating is less a character failure than a rational response to how the course is structured, and each new tool has made that response easier and cheaper to choose.

AI removed the last barrier. Every earlier method required access to something: a fraternity's filing cabinet, a Chegg subscription, an invitation to the right group chat. AI requires a browser. That is why proctoring alone won't close the gap. A proctored exam does nothing about the unsupervised practice and homework a student completes beforehand. Cheating on a practice assignment doesn't hurt a student's grade directly, but it strips out the studying, and with it the understanding the student was supposed to build for the exam and for their career.

The clearest measurement of that damage so far comes from China, where student AI adoption moved faster than almost anywhere else. Researchers tracked the study habits and exam scores of 27,000 secondary students over a six month period, comparing the roughly 80 percent who used AI with classmates who did not ("Does AI stop children from learning?", 2026). The AI users' homework scores rose 18 percent, and the time they spent per assignment fell by about a third. Their exam scores came in 20 percent below their classmates'. Homework scores, which had always been assumed to predict exam performance, now predicted the opposite: the students with the best homework grades were more likely to do worse on the exam. The one group that escaped the penalty was students who used AI as a tutor, spending as much time on assignments as non-users, instead of using it to finish faster.

So how do you convince students to put real effort into studying when the answers are one paste away?

The solution is a combination of structural changes that protect the integrity of tests and access to frequent formative assessment that prepares students for those tests. PracticeQuest was built to support both. If exams are proctored and are not drawn from a fixed test bank, there is a real incentive to learn the material rather than borrow answers, because the borrowed answers won't exist when the test arrives. And if students can be graded on their practice effort rather than accuracy, much of the incentive to get the answers fast disappears.

PracticeQuest uses dynamic question templates linked to structured content databases to generate new question variants on demand. The same system provides unlimited low-stakes practice and fresh variants for the test. Every practice attempt allows repeated self-assessment of a learning objective without repeating stale questions, and the practice works the way the tutor-mode students in the China study worked: repetition with feedback instead of answer retrieval. The practice is direct preparation for the test. With dynamic question generation, answer-sharing sites and group chats have nothing durable left to share. There are no fixed questions, no static answer key, and no two students receive the same question. PracticeQuest can't stop a student from typing a question into a chatbot; no platform can. What it removes is the thing every cheating method in this history depended on: a fixed, shareable answer. The structural half belongs to instructors and institutions: every assessment that carries an accuracy-based grade must be securely proctored. Without that piece, tests will mostly tell you who is using AI most effectively. Instructors: PracticeQuest offers free class trials for college biology courses — the course is built for you, matched to your syllabus. References

Alden, C., & Ha, A. (2020, December 16). Professors say students cheated using Chegg, following MATH 100 cheating incident. The Ubyssey. https://ubyssey.ca/news/more-students-may-have-cheated-with-chegg

Alonso, M., & Akbarzai, S. (2021, June 11). Dartmouth Medical School drops cheating sanctions against students. CNN. https://www.cnn.com/2021/06/11/us/dartmouth-medical-school-cheating-scandal

Bever, L. (2017, November 13). Dozens of Ohio State students accused of cheating ring that used group-messaging app. The Washington Post. https://www.washingtonpost.com/news/grade-point/wp/2017/11/13/dozens-of-ohio-state-students-accused-in-cheating-ring-using-group-messaging-app

Bogardus Cortez, M. (2016, December 7). Scantron: History and evolution in educational testing. EdTech Magazine. https://edtechmagazine.com/higher/article/2016/12/scantron-technology-creates-more-efficient-testing

Does AI stop children from learning? (2026, August 18). The Economist. https://www.economist.com/graphic-detail/2026/08/18/does-ai-stop-children-from-learning

Glanzer, P. L., & Cockle, T. F. (2021, July 21). How southern honor corrupted American higher education: A Christian critical history and alternative to honor codes. Christian Scholar's Review. https://christianscholars.com/how-southern-honor-corrupted-american-higher-education-a-christian-critical-history-and-alternative-to-honor-codes/

Jones, J. (2020, November 17). Georgia State students warn about cheating through GroupMe. The Signal. https://digitalcollections.library.gsu.edu/digital/api/collection/signal/id/52149/download

Pascual, M. G. (2026, June 28). Professor denounces mass AI fraud on an exam at Brown University: "Academic integrity is at risk." EL PAÍS English. https://english.elpais.com/education/2026-06-28/ai-fraud-at-brown-university-academic-integrity-is-at-risk.html

Secretary of Army will act on cadets, but Hoffmann does not confirm expulsion figure. (1976, December 16). The New York Times. https://www.nytimes.com/1976/12/16/archives/secretary-of-army-will-act-on-cadets-but-hoffmann-does-not-confirm.html

Smith, M. (2012, May 10). Test file for everyone. Inside Higher Ed. https://www.insidehighered.com/news/2012/05/11/online-test-bank-raises-copyright-issues

Watson, E. (2021, April 18). West Point cadets expelled over worst cheating scandal in 40 years. CBS News. https://www.cbsnews.com/news/west-point-cheating-scandal-cadets-expelled

Whitford, E. (2026a, May 15). Princeton introduces proctoring, changing honor code. Inside Higher Ed. https://www.insidehighered.com/news/faculty/learning-assessment/2026/05/15/princeton-introduces-proctoring-changing-honor-code

Whitford, E. (2026b, July 8). Brown professor suspects most of his class used AI to cheat. Inside Higher Ed. https://www.insidehighered.com/news/faculty/learning-assessment/2026/07/08/brown-professor-suspects-most-his-class-used-ai-cheat