The case that changed the debate

The case emerged after the national exam for accountants: according to The Korea Herald, the first-place finisher said he had photographed the questions with a microcamera, sent the images to an AI tool, and read the generated answers on his phone. The method required an internet connection, an element that paved the way for a possible technical intervention ahead of the College Scholastic Ability Test, the Suneung, scheduled for Thursday with more than 550,000 candidates.

The education authorities were still looking for solutions two days before the test. The option under study was jamming wireless signals at exam centers to prevent devices from transmitting photos or receiving answers.

The legal constraint and practical concerns

On Monday, a briefing by the Ministry of Education confirmed that signal jamming will not be used for this edition of the CSAT. The measure is strongly limited by the Radio Waves Act, which allows signal-blocking devices only in narrow circumstances related to public safety, such as certain security, counterterrorism and military operations. The ministry also stated that a revision of the law for broader use would require careful assessment.

A ministry official explained that schools are located near residential areas: it must be verified whether the measure could interfere with emergency responses in exam centers and whether it could have unintended effects on the surrounding residential areas. The ministry cited legal and practical obstacles to other technological solutions as well: physical searches of students raise legal issues and can cause psychological distress, while video recording during tests would require the consent of all candidates, making general application difficult.

The chosen approach and foreign models

The ministry’s approach is therefore based on the fact that communication devices have physical characteristics that proctors can recognize through visual inspections; blocking public Wi-Fi at exam venues during the test is also being considered. Criticism of the government has turned the spotlight on strategies adopted elsewhere. In China, where the national university entrance exam carries weight comparable to the CSAT, AI-based surveillance systems have been introduced that analyze camera images in real time and flag suspicious behavior to proctors, such as repeatedly turning one’s head or hiding one’s hands under the desk.

According to the newspaper, major Chinese AI services, including Alibaba’s Qwen and DeepSeek, suspended during the exam period image-recognition functions capable of solving questions from photographs. Such measures would be easier to coordinate in China, where access to foreign AI services such as ChatGPT is limited. India has adopted stricter controls, including surveillance footage and radio-frequency monitoring in some exams, and has toughened the penalties under its law on preventing unfair means in public examinations: amendments in force since July provide for five to ten years in prison and fines of up to 5 million rupees for those involved in fraud, and seven to ten years with fines of at least 100 million rupees in cases of organized fraud.

What remains open

In South Korea, the ruling Democratic Party is considering a law that would extend to ten years the period during which those responsible for fraud can be excluded from exams. Critics, however, believe such measures would come too late relative to the immediate threat to this year’s CSAT, and some have called for postponing the test until more effective safeguards are introduced. Prime Minister Han Seong-sook announced that proctors will check every pair of glasses and every watch worn by candidates. The practical effectiveness of these checks and the possible blocking of public Wi-Fi remain to be verified, measures that would not in any case prevent connection via mobile networks.

Source and attribution: this original summary was prepared by AICronaca based on the report by The Korea Herald ↗. Statements are attributed to the source; any information limitations: The material is based on a single article from The Korea Herald and does not contain data on the number of confirmed frauds, on the results of the announced measures, or independent confirmation of the statements by the ministry and the prime minister. Information on foreign models is reported as statements from the source..