The government has begun a nationwide review of 528 national qualification examinations after a top scorer in the tax accountant examination admitted using generative artificial intelligence to cheat. The case has widened concerns beyond one test. Candidates and professional groups are questioning whether current controls can protect the fairness of examinations taken by millions of people each year.
The review covers examinations administered this year, including tests for professional and technical qualifications. Reports have described concerns involving labor attorney, fire equipment engineer, customs broker, certified appraiser and civil service examinations, as well as university entrance and recruitment tests. Some of these cases remain allegations, not confirmed cheating. The governmentโs investigation therefore has a wider remit than the individual case that prompted the latest scrutiny.
The suspected methods illustrate why conventional supervision is under pressure. Reports have described the use of miniature cameras to transmit question images to a smartphone and tiny earpieces to receive answers generated by an AI system. Experts cited in the coverage proposed checking glasses in advance, providing supervisors with metal detectors and using light to identify cameras embedded in smart glasses. These proposals address equipment that can be concealed or used without the visible behavior that traditional invigilation is designed to detect.
The controversy also exposes a difficult limit on any investigation conducted after results are released. If other candidates used undiscovered assistance and passed, authorities may not be able to identify every case or restore equal treatment through a complete retrospective review. Immediate controls at examination sites and longer-term changes to testing procedures will therefore have to proceed together. Until the government reports what the 528-examination review can establish, doubts about individual tests should remain separate from confirmed findings, while the admitted AI cheating case continues to shape confidence in the system.
