At least 50 students in an advanced economics course at Brown University used artificial intelligence to cheat on a take-home midterm exam, according to professor Roberto Serrano, who calls the incident the largest known academic fraud case in the Ivy League. The university's muted response has left Serrano demanding a public reckoning over how AI threatens the credibility of higher education.
How the fraud was uncovered
Serrano, the Harrison S. Kravis University Professor of Economics, taught ECON 1170, a mathematical economics course that normally enrolls fewer than 30 students. This semester, 86 signed up after he announced that both the midterm and final would be take-home, closed-book exams-a format he chose to make the test more challenging while giving students unlimited time.
The March 5 midterm produced extraordinary results: an average score of 96 out of 100, with 40 students earning perfect marks. Teaching assistants who graded the exams flagged irregularities. "Some answers contained unusual passages that coincided with results obtained after running the questions through ChatGPT," Serrano said.
Serrano did not void the midterm but warned students that the in-person final would count for 50% of the grade. He also said that if the grade distribution shifted dramatically, only the final exam would be used. The average score on the final plunged to 48. Of the 89 students who took the midterm, 59 showed up for the final; among the 27 who stayed away, 22 had scored a perfect 100 on the earlier test.
"The empirical evidence of fraud is overwhelming," Serrano said. He plans to eliminate take-home exams and stop counting weekly problem sets toward final grades, since those can be completed with AI.
A campus tragedy and a lenient policy
Serrano's decision to offer take-home exams this semester was shaped by a shooting on Brown's campus in December 2024. A former PhD student opened fire during a review session, killing one student and injuring nine others. Two of the injured were enrolled in Serrano's class. The student who died, Ella Cook, had recently met with Serrano to discuss enrolling in his Intermediate Microeconomics course and asked him to be her advisor.
"I was in a really bad place mentally for a while," Serrano said. "After what happened, it occurred to me that exams could be take-home in order to make life a little easier for students. Many of them still feel anxiety when they are on campus because of what happened in December."
When Serrano reported the cheating to senior administrators, he received what he described as a cold reaction. The president did not respond, and the dean only commented after Serrano brought the case to the Academic Code Committee, calling the incident "a wake-up call." Serrano believes that falls short. "Academic integrity is a value worth defending. The faculty cannot be left on its own in a battle that is decisive if we want to preserve the future of higher education," he said.
He also worries that the university's reluctance to act stems from its reliance on donations from wealthy families whose children often attend Brown. "This means that the kids always get the benefit of the doubt; I've seen it on other occasions," he said.
The broader AI cheating problem
Serrano's case is not isolated. Princeton University recently ended a 133-year-old honor code that allowed professors to leave the room during exams, requiring in-person proctoring for the first time since 1893. The shift reflects how AI tools have made cheating easier to hide and harder to police.
"A.I. has made deception easier and more remunerative than ever before," journalist Theo Baker wrote in The New York Times, quoting a recent Stanford graduate who said, "I don't know a single person who hasn't used A.I. to get through some assignment in college."
Serrano argues that universities cannot sweep these incidents aside. He wants institutions to publicly acknowledge the scale of the problem and launch a serious debate about AI for Education and academic integrity. "If we no longer defend truth and decency and honesty, then what kind of credibility are we going to have as academics?" he said.
Why this matters for educators
The Brown case signals that even at elite institutions, existing honor codes and take-home assessments are failing against AI. Faculty need institutional backing to redesign exams that test reasoning beyond what a chatbot can generate. That means shifting toward in-person assessments, oral defenses, or problem sets that require students to apply concepts to novel scenarios. It also means pushing administrations to create clear, enforceable AI policies instead of leaving individual instructors to fight cheating alone. Without that support, the trust that underpins grading and credentials will continue to erode.
Your membership also unlocks:
