South Korea tightens CSAT checks over AI cheating fears | Magica
South Korea tightens AI cheating checks without device detectors
Editorial Team
••📖6 min read
South Korea is strengthening device checks and considering public Wi-Fi shutdowns before its November 19 college entrance exam, after a tax-accountant exam top scorer admitted using AI to cheat. The response still depends heavily on human screening, with no electronic device detectors planned for this year's test.
South Korea plans stricter device checks before its November 19 college entrance exam; public Wi-Fi shutdowns remain under review.
Electronic device detectors will not be used this year, leaving proctors and specialist staff central to prevention.
The government is also reviewing national qualification exams and pursuing sanctions over a tax-accountant exam cheating confession.
South Korea is tightening checks for concealed electronic devices ahead of its November 19 college entrance exam after a tax-accountant exam top scorer admitted using artificial intelligence to cheat. But electronic device detectors will not be introduced for Thursday’s test, and a proposal to cut public Wi-Fi at exam sites remains under review.
The College Scholastic Ability Test, or CSAT, is South Korea’s annual national university entrance examination, known in Korean as Suneung. The test helps determine university access and future career opportunities; public transport, flights and even military drills are adjusted to minimize disturbances, according to the report on national preparations. The immediate challenge is to detect concealed devices without disrupting legitimate candidates.
Smart-glasses training preceded the tax-accountant exam confession. An interministerial response meeting followed on November 16. Graphic: Magica. Source: Korea JoongAng Daily.
A confession accelerates an existing crackdown
The immediate trigger was the November 15 admission by the top scorer in this year's certified tax accountant examination. After weeks of suspicion, he said he used a miniature camera, a smartphone and generative AI in both rounds of the test. He missed only one of 160 questions in the first round and earned high marks in the second, according to a report on the confession and ensuing police investigation.
That admission concerns a separate qualification exam; it does not establish that the same method has succeeded in the CSAT. Nor does the available reporting establish how widespread undetected AI cheating is.
Russia’s objections to observer participation stalled the opening of a Geneva weapons conference, where governments must decide whether to authorize negotiations on autonomous weapons and whether any resulting rules should be binding.
Editorial Team
There were already reasons for concern. One middle school student and three high school students were caught using AI glasses during school exams this year, while devices were also detected during separate university application tests, the same report said. The report said some AI glasses had been found capable of scanning tests and producing answers within seconds using generative AI. The problem facing supervisors therefore extends beyond recognizing a conventional phone on a desk.
At an emergency meeting on November 15, the presidential office ordered an examination of suspected AI cheating across national qualification tests. Its statement also described preparations for checking all CSAT candidates, assigning personnel to identify AI devices at each testing site and strengthening training for candidates and supervisors.
The office called for using available methods, including thermal cameras, while explicitly warning against harming candidates wearing ordinary glasses or medical devices. It also called for examining the distribution of equipment such as miniature cameras and encouraging tip-offs. Investigators would form dedicated teams for serious cases involving new methods or organized cheating, the statement said. That instruction describes the requested response, rather than evidence that thermal screening has been deployed nationwide.
Sejong’s 16 designated specialists are existing education-office supervisors. Two proctors remain in each classroom to make the first assessment. Graphic: Magica. Source: Yonhap News Agency.
More scrutiny, with limits on detection
The government's November 16 interministerial meeting put the Education Ministry in charge of an emergency response team for the exam, which serves the 2027 academic year. Candidates are to surrender all smart devices, and checks for possession will continue during the test, according to the official account of the meeting.
The Ministry of Science and ICT was asked to examine blocking public Wi-Fi at all testing sites during examination hours in consultation with relevant agencies. The distinction matters: the government is reviewing a particular network restriction, not reporting that every means of outside communication has been disabled.
The response spans 1,312 test sites nationwide. Personnel specializing in detecting AI glasses are to be dispatched, but electronic device detectors under consideration for future exams will not be introduced this year, according to the report on national preparations. A ministry official said physical searches and metal detectors would be difficult to implement and could conflict with maintaining a stable testing environment.
Sejong's arrangements show what the additional supervision means locally. Its education office will assign one smart-glasses specialist to each of 16 testing schools. All 16 are existing education-office supervisors given a new responsibility, rather than additional recruits, the office explained in a report on its preparations.
According to the education office’s account, the usual two proctors will remain in each classroom. They will make the first assessment of a suspicious situation, consult the designated specialist next and escalate difficult judgments to the education office. The specialist provides another level of judgment, but does not replace the initial classroom check.
The Korean Teachers and Education Workers Union has challenged reliance on individual proctors' vigilance and called for discussion of how assessment should change as AI becomes better at supplying answers. Its commentary is described in the national preparations report.
Chun Ha-ram, floor leader of the minor New Reform Party, has called for systematic preventive measures even if that requires postponing the exam, the same report said. The government’s announced preparations continue to target November 19.
Investigations and separate penalties
Officials plan to keep examining suspected cheating after candidates leave the room. The government will operate a reporting center, investigate suspicious cases with police and check anomalous results, including implausibly high scores, according to a report detailing the enforcement plans. Sanctions are to follow confirmed misconduct; an unusual score is a reason to investigate, not itself proof of cheating.
Candidates caught with smart devices during the CSAT face invalidation of their results and exclusion from next year's exam. The government is moving to extend that exclusion to three years, according to the national preparations report. The Education Ministry also plans to seek criminal punishment for offenders. Separately, the prime minister's office ordered the start of proceedings to invalidate the tax-accountant exam top scorer's result and suspend his eligibility for that examination for five years. Those are distinct sanctions for different examinations.
What remains to be decided
The longer-term response reaches beyond university admissions. The government ordered a cross-ministry task force, while the Ministry of Employment and Labor and the Human Resources Development Service of Korea—the body that runs national certification exams—are to review the qualification system. The review includes the necessity and effectiveness of existing exams, alongside proposals such as certified testing venues and smart screening gates. The official meeting account also calls for exploring cooperation with global AI companies, without specifying a mechanism or agreement, and considering changes to laws and rules to strengthen sanctions. It also calls for a standing online and offline reporting system for cheating. The announcement provides no detection-performance results for the proposed screening equipment.
For Thursday's test, the unresolved questions are more immediate: whether the public Wi-Fi proposal will be adopted, how reliably staff can distinguish concealed electronics from permitted items, and how checks can be conducted without distracting legitimate candidates. The announced measures establish procedures and penalties. Their effectiveness at preventing AI-assisted cheating remains to be demonstrated.
Apple’s iOS 27.3 beta ties higher allowances for server-based AI to existing iCloud+ plans. Basic Siri commands remain free, but undisclosed daily limits leave the value of paid access uncertain.
The partners plan to combine IBM quantum processors, NVIDIA computing and Fermilab physics expertise to study modular quantum systems. A separate decoder release promises faster error-correction research, with important limits on its performance comparisons.
Federal Reserve Chairman Kevin Warsh outlined AI scenarios spanning shared prosperity, job displacement and an investment downturn. The exercise offers no forecast or rate guidance as colleagues worry that the infrastructure buildout is adding to inflation.
OpenAI is offering approved organizations a free classifier for suspected child sexual abuse images, including AI-generated material. Early results suggest less review work, but existing detection alternatives and the need for human review qualify the launch.
Nvidia plans scientific computing racks for early 2027, extending Vera beyond its initial AI deployments. Its 2.4× performance-per-watt claim comes from one airflow simulation against AMD Turin; performance against newer rivals and delivered costs remain unresolved.
OpenAI’s CEO expressed empathy for Nvidia founder Jensen Huang but withheld judgment on reinstatement. His confidence in continued AI financing comes with a direct commercial stake and unresolved deployment risks.
Google’s 8,192-chip Ironwood submission completed the MLPerf DeepSeek-V3 benchmark in 1.70 minutes, close to NVIDIA’s 1.53 minutes at the same accelerator count. Google’s scaling data also shows the extra chip-time required for its fastest result, leaving cost and energy comparisons unresolved.
TensorWave will supply dedicated AMD infrastructure under a six-year OpenAI agreement, with delivery expected from 2027. Its operating record and financial backing sit alongside an undisclosed contract price and a recently terminated site lease.
California alleges OpenAI endangered the public and misled consumers by testing powerful AI models without adequate controls. The lawsuit follows a Hugging Face intrusion that OpenAI’s own account traced to containment failures, missing monitors and mishandled warning signs.
Perplexity has launched Frontier Research to develop AI that can formulate questions, test hypotheses and produce interpretable discoveries. The company promises more than 10,000 GPUs, but existing automated-research systems show why computing resources and generated papers alone cannot establish scientific value.
The new lab led by Stanford professors Yejin Choi and Carlos Guestrin has backing from Nvidia, Cisco and Snowflake. It promises systems that improve on company work, but research benchmarks leave production costs and customer control unproven.
NVIDIA announced a Rubin HPC GPU with more than 50,000 CUDA cores, sparse data acceleration and seventh-generation NVLink. The design targets simulations used by scientists and AI agents, but missing performance, power and pricing details leave its practical advantage unproven.
OpenAI will open 50,000 new ChatGPT access grants on November 23, covering up to 150,000 researchers and collaborators. New recipients get Plus instead of Pro, with first-come selection replacing a lottery. Its pilot study reports perceived time savings but does not establish faster discovery.
OpenAI reports production healthcare workflows at CVS and Cigna and stronger computer-use evaluations. The results show progress across disconnected software, but leave complete-task reliability and total deployment costs unresolved.
President Donald Trump supported a return for Nvidia co-founder Jensen Huang, but the company has installed interim leadership and begun a successor search amid scrutiny of his conduct and alleged Chinese intelligence ties.
Intel disclosed the bandwidth target for its next Xeon server processors at SC26, alongside previously announced core and cache limits. Application performance, power requirements and launch timing remain unresolved.
Mercedes-Benz plans California early access to city-driving and hands-free highway assistance in December, ahead of a 2027 U.S. rollout. Mixed road tests and unannounced subscription prices leave buyers waiting for evidence of everyday value.
Researchers estimate that 403 large U.S. data centers consumed 81.8 TWh in a year. Their regional-grid analysis reveals substantial fossil-fuel dependence, but cannot isolate AI demand or fully expose the underlying facility data.
HSBC lowered its Zhipu share-price target to HK$500 and delayed forecast adjusted breakeven by two years. Improving API margins and substantial fundraising cushion a business whose gross profit still covers little of its research bill.