Brown Professor Says Most of His Class Cheated With ChatGPT

An economics professor at Brown University says most of his class cheated with ChatGPT this spring, and the way he found out is the part every student should read.
Roberto Serrano teaches an advanced course on welfare economics. It usually draws about 30 students. This spring, 86 signed up. Serrano thinks he knows why: he had promised take-home exams.
The numbers that gave it away
The take-home midterm came back with an average of 96 out of 100. Forty students scored a perfect 100. In previous years, the average on that same midterm ran between 65 and 80.
Serrano and his graders ran the exam questions through ChatGPT. The answers that came back mirrored what his students had turned in, down to the reasoning style.
Students were reaching for contradiction arguments when a direct proof was the obvious move, which is exactly what the chatbot did. He called the writing "kind of correct, but very off and with a very convoluted style."
Then he announced that the final would be in person, on paper, in a room. What happened next is the part that spread:
18 students dropped the course after the announcement.
9 more never showed up to take the final.
Of the 27 students who skipped that in-person exam, 22 had scored a perfect 100 on the take-home midterm.
The in-person final averaged 48.6. In the history of the course, it had never dropped below 65.
Nineteen students failed. Serrano lowered the passing threshold from 50 to 40 to soften it. His conclusion, in his own words: "The empirical evidence of fraud is overwhelming."
Why the exam was take-home in the first place
This detail matters, because it is easy to read the story as a professor being naive. He was not. In December 2025, a mass shooting on the Brown campus killed two students. One of them, Ella Cook, had recently asked Serrano to be her academic advisor.
He moved to take-home exams so that students who were still shaken would not have to sit in a classroom to be graded. "I wanted to remove the stress," he told Fortune. He now says the take-home exam "is a thing of the past."
Brown has not done much about it
Serrano submitted his findings in May. He says the dean and the provost did not respond, and that the university committee that handles academic integrity asked him to file a separate individual complaint against every suspected student. He went public in the Spanish newspaper El Pais in late June.
A Brown spokesperson, Brian Clark, told Inside Higher Ed that the university "treats every allegation of academic integrity with the utmost seriousness," but that Serrano "has not provided the necessary details" to move forward. So far, no student has been penalized.

Source: Business Insider
What this changes for you
The take-home exam is being pulled back, and not just at Brown. Professors are moving to in-person tests, oral exams, and proctored formats.
Some of the new setups are strange: students in one UCLA class were told to keep a mirror large enough to reflect their whole desk, with the laptop camera on, so the professor could watch them work.
Two practical takeaways. First, plan for closed-book, in-person assessment as the default again, which means studying for real rather than counting on an open laptop.
Second, protect yourself if you did nothing wrong. AI detectors produce false positives, and researchers have found they misfire most often on writing by students whose first language is not English.
That is a real risk for honest students. Write in a document with version history turned on, keep your drafts and your notes, and you have something to show if you are ever accused. We covered how unreliable this detection is in our piece on whether colleges check applications for AI.
And if your school hands you an AI tool, learn what you are allowed to do with it. Many campuses now pay for one on your behalf, with rules attached that most students never read.