Submit incident
Documented

Half a Yonsei Class Cheated on an AI Exam Because No Policy Said They Couldn't

January 1, 2025
Curated by Team Raidu · Reviewed by Shiva Ganesh
aiaaic:AIAAIC2115View source ↗
LinkedInX

What happened

In October 2025, about 600 students at Yonsei University sat for an online midterm in a course called "Natural Language Processing & ChatGPT." The subject matter was generative AI. The exam format was online. When the professor reviewed submissions, a significant number of students had apparently used the very tools the course was teaching them about to produce their answers.

The university had put what looked like serious safeguards in place: students were required to record their screen, their hands, and their face for the duration of the test. Despite those measures, many students allegedly manipulated their camera angles and ran their answers through AI tools in parallel windows, evading detection in real time. The professor announced that every student caught would receive a zero for the exam and face potential suspension.

What turned this from a standard academic discipline matter into a public controversy was an anonymous poll on a student community board. Of 353 respondents in a survey titled "Let's Vote Honestly," 190 admitted to cheating, suggesting more than half the class had engaged in misconduct. That number was self-reported and voluntary, which means the actual share could be higher. The poll was not a formal investigation; it was students telling each other, in a space the professor could not see, what had actually happened.

The incident exposed a mismatch that Korean universities had not yet resolved by fall 2025. Many institutions, including top-ranked ones, had not adopted clear guidelines for generative AI use in coursework or exams. An online exam is particularly ill-suited to detection when a student can open a second window and query a language model in seconds. Students reported feeling that not using AI put them at a disadvantage, a pressure the absence of enforced rules did nothing to relieve. Legacy assessment formats built for a pre-generative-AI world were still in use unchanged.

What the Yonsei exam incident shows is how quickly an integrity system collapses when there is no way to verify what actually produced a submitted answer. The professor could catch some students by reviewing recording footage, but there was no mechanism for establishing, at the moment of submission, whether the work was the student's own. A provable record of what a system did, when it ran, and whether a human or an automated tool generated the output would close exactly that gap, without requiring anyone to monitor six hundred simultaneous video feeds frame by frame.

Reported impact

Affected parties
Not publicly disclosed
Harm type
Not publicly disclosed
Scale
Not publicly disclosed
Financial impact
Not publicly disclosed
Regulatory action
Not publicly disclosed

Classification

Organization
Not publicly disclosed
AI system
Not publicly disclosed
Industry
Not publicly disclosed
Country
Not publicly disclosed
Provider
Not publicly disclosed
Incident type
Not publicly disclosed

Relevant governance controls

Governance control mapping is not available for this record.

  • No controls mappedNot publicly disclosed

Control mapping is analytical. It does not state that any control would have prevented the incident.

Sources and evidence

This record was researched and written by the Index. The event is also catalogued in the following database, which is listed for cross-reference.

AIAAIC Repository
Also catalogued in
Half a Yonsei Class Cheated on an AI Exam Because No Policy Said They Couldn't
2025