Hundreds of students use ChatGPT in Yonsei University online exam

What happened

An online midterm exam held on 15 October in a class titled 'Natural Language Processing (NLP) and ChatGPT' with approximately 600 students at Yonsei University's Sinchon campus saw widespread cheating. Students bypassed proctoring measures by adjusting camera angles to create blind spots, overlaying multiple programs on screen, and capturing exam questions.

Many students used ChatGPT and other AI systems during the exam. In an informal poll on the student community platform Everytime with 387 responses, 211 students reported cheating and 176 reported solving the exam honestly. The professor announced that admitted cheaters would receive zero points on the midterm, while students who deny cheating face expulsion.

Reported impact

By November, over 40 students had admitted to cheating. The university stated that 10 additional students suspected of cheating had not confessed. The midterm exam scores of all admitted cheaters were set to zero.

Index analysis

The incident reflects inadequate proctoring in remote exams and insufficient institutional governance on AI use in academic settings. According to research cited in reports, 71.1% of 131 surveyed universities nationwide lacked generative AI guidelines, and 91.7% of surveyed students had used AI in coursework or research.

Governance controls

The course required students to submit video footage showing their computer screen, hands, and face throughout the exam. However, these measures proved insufficient to prevent cheating.

Classification

Organization

Yonsei University (victim)

Category

Fraud, scams and manipulation

Industry

Education and training

Country

South Korea

Harm status

Actual harm

Intent

Intentional

Timing

Post-deployment

Harm types

Reputational

AI systems

ChatGPT

Sources and evidence

Related incidents