evidoria

← Back to browse

Good practice Imported

Brown University's AI Cheating Scandal — A Cautionary Case in Undetected GenAI Use on Take-Home Exams

United States of America · Providence · See the United States of America profile

Evidence: Descriptive / self-reported Top 100% 7/100 · Ask Evidence Copilot about this practice

After Brown University economics professor Roberto Serrano moved a midterm to a take-home format, grades spiked implausibly — 40 of 86 students scored a perfect mark — exposing how easily generative AI evades detection on remote assessments.

96
Take-home midterm class average (Spring 2026)
40 of 86
Students scoring a perfect mark on the take-home midterm (Spring 2026)
65–80
Typical class average on prior in-person midterms
48.6
Class average on the subsequent in-person, closed-book final exam (Spring 2026)
59
Students who sat the in-person final exam (of 86 enrolled) (Spring 2026)
Brown University's AI Cheating Scandal — A Cautionary Case in Undetected GenAI Use on Take-Home Exams

Details

Promoter
Brown University, Department of Economics (ECON 1170)
Period
Spring 2026 semester; publicly reported June–July 2026
Keywords
higher education, academic integrity, assessment design

Context

In spring 2026, Brown University economics professor Roberto Serrano moved the midterm exam for course ECON 1170 to a take-home, online format, after enrolment jumped to 86 students from a previous record of 30. The take-home midterm average came out at 96, far above the typical 65–80 range for prior in-person versions, with 40 of the 86 students scoring a perfect mark and many answers showing the hallmarks of ChatGPT-style output.

Results

Serrano said he suspected at least 50 students of using generative AI on the exam, and that the university was slow to open a formal investigation until the case drew public attention. When he reverted the final exam to an in-person, closed-book format, only 59 of the enrolled students sat it and the class average dropped to 48.6 — a contrast independent journalists have treated as strong circumstantial evidence of undetected AI-assisted cheating on the take-home version. Brown has not published its own official statistics; the figures come from the professor's account as corroborated by multiple news outlets.

Implementation
Indicative cost
Low (< €50k)
Time to results
Short (< 1 year)
Staffing & skills
course instructor (Prof. Roberto Serrano), university academic-integrity investigators

Conditions for success

  • closed-book/in-person format or verified proctoring for high-stakes take-home exams
  • prompt formal investigation when grade patterns look implausible

Common failure modes

  • take-home, unproctored format used for a high-stakes exam
  • slow institutional response until the case drew public attention

Commonly funded by

Own resources / municipal budget

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Attachments

Similar practices you may find useful