Online finals at Korean universities again marred by cheating

The entrance to Seoul National University's Gwanak Campus in southern Seoul / Korea Times file

Mass cheating in online exams has resurfaced during finals at Korean universities, exposing institutions’ incomplete preparedness and ongoing struggles to establish effective online test controls and guidelines for students' artificial intelligence (AI) use.

According to Seoul National University, Sunday, the results of a final exam for a general education course offered by its College of Natural Sciences were invalidated after signs of cheating were detected among nearly half of the 36 students enrolled.

The course was an online distance-learning class designed for students on leave for mandatory military service, with both lectures and exams conducted remotely.

To prevent misconduct, the exam system was set to record activity whenever test-takers opened any windows other than the exam screen. A subsequent review by a teaching assistant later found such logs for nearly half of the students, leading the university to invalidate the exam results.

However, it has been difficult to conclusively establish misconduct because the logs do not indicate which screens were accessed. As a result, the professor overseeing the course invalidated the exam results and assigned alternative coursework, rather than imposing disciplinary penalties on students.

The professor noted that while definitive proof was lacking, the scale of the suspicious activity strongly hinted at academic misconduct.

Seoul National University said it was drafting "institutionwide steps" in response to what it described as "widespread cheating."

Reportedly, discussions are underway about making in-person exams the default, while turning to alternatives for online assessments — including open-book formats, redesigned questions and assignment-based evaluations — when remote testing is unavoidable.

The university said it is also moving to introduce "campus-wide guidelines on AI use."

Under the proposed framework, instructors would be required to clearly outline their AI policy in the course syllabus, enabling students to understand the expectations in advance. The draft guidelines are currently being reviewed through consultations with faculty and students, according to the university.

The university had already been rocked by an AI-related cheating case during the October midterm exam for an introductory statistics laboratory course, underscoring the urgency of stronger measures to curb academic misconduct.

A sign of the school stands at the gate of Yonsei University’s main campus in Seoul, Dec. 1, 2024. Newsis

The wave of mass cheating is not confined to Seoul National University, with similar cases resurfacing at other elite universities in Korea following AI-related cheating scandals during midterm season.

Earlier this month, Yonsei University uncovered another case of mass cheating during the October midterm exam in the same course that had been at the center of a large-scale cheating scandal involving the use of artificial intelligence.

It occurred during an online quiz taken by roughly 200 students, where some participants were found to have exchanged questions and answers through anonymous online chat rooms. Some were also reported to have shared exam content in real time using collaborative platforms such as Google Docs.

The course is delivered through prerecorded online lectures and had already drawn allegations of academic misconduct during the midterm exam. In response, the university moved the final exam to an in-person format, while online quizzes continued to be administered remotely.

Yonsei University has had guidelines on the use of artificial intelligence in place since last year and updated them in September. But the framework remains broad and largely advisory rather than enforceable, leaving individual courses to apply differing standards for the use of such tools.

Critics say universities have largely relied on ad hoc responses — including invalidating exams — when cheating cases surface, an approach they warn could deepen mistrust between students and administrators instead of addressing the underlying causes of academic misconduct.