South Korea probes AI cheating in tax-accountant exam | Magica
South Korea opens AI exam-cheating probe and widens qualification review
Editorial Team
••📖7 min read
Police are investigating a top tax-accountant exam scorer's admission of AI-assisted cheating, while the government reviews national qualification tests held in 2026 and prepares stronger controls.
Police assigned a tax-accountant candidate's cheating confession to a specialist metropolitan investigation unit.
The government ordered a review of national qualification exams held in 2026, using the suspected pattern to look for other cases.
The admission raises questions about exam supervision; it does not establish the prevalence of AI-assisted fraud.
South Korean police opened an investigation on November 15 after Han Won-seok, the top scorer in this year's certified tax-accountant examination, admitted using a miniature camera, a smartphone and generative AI during both rounds. The government separately announced a review extending across national qualification examinations held in 2026.
The confession followed the qualification results and came days before the scheduled nationwide university entrance examination. Graphic: Magica. Source: P&P News.
A confession puts exam supervision under scrutiny
The Seoul Metropolitan Police Agency said it received a voluntary surrender statement through the candidate's lawyer that morning and immediately assigned the case to its Metropolitan Investigation Unit. The unit handles major violent and economic crimes; assigning this examination case there rather than to a local station was unusual, according to the report on the police response.
Han appeared on his lawyer's YouTube channel and acknowledged cheating in the April first-round and July second-round examinations. In his account of his circumstances, he described leaving a job and then struggling with an unsuccessful business before seeking the qualification.
He said he photographed questions with a miniature camera, transmitted them to generative AI and checked answers on a smartphone. He also said some equipment had been damaged but that he would submit the remaining devices and evidence to investigators. The available accounts do not identify the AI model or independently establish how the equipment operated throughout the tests.
His unusually high scores had prompted online suspicions about smart glasses, which he initially denied, according to . His subsequent description concerned a camera and phone. That distinction matters to establishing which devices escaped detection.
An early tally puts Japanese listed companies’ first-half net profit growth at 43.2%, led by manufacturers. Filings show AI component demand alongside currency benefits and non-AI earnings gains, while procurement delays and canceled infrastructure reservations expose the costs of expansion.
The Human Resources Development Service of Korea, or HRDK, administers the tax-accountant examination. It said two supervisors were present in the candidate's examination room. His seat was in the centre of the second row for the first round and the centre of the front row for the second, according to the report citing the administrator. If his account is substantiated, proximity to supervisors did not prevent the alleged conduct.
HRDK had previously said its own inquiry found no evidence supporting the allegations. On November 13, it said the four supervisors across the two rounds had reported no suspicious behaviour or unusual circumstances. The account of that inquiry also said the classrooms had no CCTV, limiting the ability to reconstruct events from video. A failure to find evidence was therefore not an independent verification that no cheating occurred.
Reported second-round pass rates were 10.48% and 10.92%. These announced outcomes do not measure cheating. Graphic: Magica. Source: P&P News.
A scarce qualification, with consequences beyond one result
Han reportedly scored 397.5 out of 400 in the first round, missing only one of 160 questions. He also received high marks in the essay-based second round. Those results attracted scrutiny after he discussed his scores in a media interview, according to the account of the controversy.
The examination offers a tightly contested route into tax practice. An October 28 report on HRDK's results announcement said 722 of the 6,606 candidates who sat the 2026 second-round examination passed. That is about 10.9%, calculated from the reported counts, compared with about 10.5% the preceding year, when 728 of 6,943 candidates passed. The proportion rose because the number sitting fell faster than the number passing. These are announced examination outcomes, not measures of cheating.
Passing is also not the final step into practice. Successful candidates must complete practical training before registering as tax accountants and beginning tax-agent work. The Korean Association of Certified Public Tax Accountants operates that training; this cohort's programme is scheduled to begin on November 19 and run until June 4, 2027, combining classes, video instruction and a placement.
Han also said he had used the same method in a certified public accountant examination. That further admission requires verification. He attributed his cheating to financial difficulties and anxiety about starting again in his mid-30s, as recorded in the report of his statement.
The Ministry of Employment and Labor and HRDK announced that they would immediately examine all national qualification tests conducted in 2026, using patterns from this suspected case to identify possible AI-assisted cheating. Their emergency-meeting statement also called for online and offline reporting centres, cooperation with police and prompt action to invalidate the candidate's examination and cancel his pass result, with a five-year examination ban.
The response had begun before the confession. The government policy coordination office said on November 13 that it would thoroughly investigate the tax-exam allegations and inspect the wider national-examination management system, according to the account of the earlier intervention. Sunday's announcement specifies an immediate review using this case's suspected pattern; it supplies no count of verified violations across those examinations.
There is relevant precedent. HRDK said an inspector assigned to detect AI-device misuse caught a candidate wearing smart glasses during a gas-engineer certification examination in October, according to the report citing the administrator. Separate education-ministry data collected from schools between October 27 and November 3 recorded four smart-glasses cheating cases in middle and high schools during the year. Two college applicants were also caught using AI glasses at entrance examinations at two Seoul universities in October, according to the report describing those incidents. These are distinct settings and detection counts, not a national prevalence estimate.
The labor ministry distinguished enforcement of existing controls from additional equipment. Its emergency statement called for rigorous implementation of measures already in place: stronger warnings about prohibited smart devices, longer supervisor training and expanded deployment of dedicated inspectors alongside ordinary supervisors. It also called for prompt introduction of additional measures, including radio-signal detectors under development by the Korea Communications Agency. The statement provides no deployment count, budget or measured detection performance for those devices.
What investigators and exam authorities must establish next
Police must test the confession against the remaining equipment and other evidence, including the further admission concerning the public-accountant examination. For exam authorities, the unresolved questions are whether existing supervision procedures were followed and whether the suspected method appears elsewhere.
The immediate deadline is the nationwide university entrance examination, known as the Suneung, scheduled for November 19. The presidential office said it was treating the developments seriously ahead of that test, according to the report on its response. A further interministerial meeting was scheduled for November 16, while the labor ministry and HRDK committed in their emergency statement to a joint task force and comprehensive measures against qualification-exam cheating by year-end.
The year-end plan must address the gap between controls on paper and detection in an examination room. Neither the admission nor the announcement establishes how widely the method was used or how reliably the proposed detectors would catch it.
TSMC plans to back promising chip startups through a new investment arm, alongside separate commitments to OpenAI and Anthropic. The strategy reaches toward future customers and the AI developers shaping chip demand, but the investment rights remain unclear.
Nidus has raised $30 million across pre-seed and seed rounds to automate wire processing in existing U.S. defense factories. Its founders bring manufacturing and government-deployment experience, but customer access and proprietary data still need to translate into measurable production gains.
Nebuly is funding US and European expansion for its AI business-value platform. Adjustable assumptions and private deployment are part of its pitch, but rivals offer similar measurement tools and verified customer returns remain undisclosed.
Han Won-seok says he used a miniature camera, generative AI and a smartphone in tax accountant and CPA exams. His confession puts scrutiny on exam security after proctors reported no suspicious behavior.
Thousands protested against proposed AI data centres in Scotland as ministers prepare new planning guidance, putting public participation and the scale of electricity demand at the centre of the dispute.
Uber is notifying drivers after a breach at AI infrastructure provider Fireworks exposed personal and driver’s license information. Three reported state counts add up to about 7,870 people. Fireworks has since disclosed wider data exposure, but the precise connection used to reach Uber’s records remains unexplained.
Jensen Huang’s reported resignation puts Nvidia’s financing and advocacy for American open models under scrutiny. Researcher Nathan Lambert warns of a loss of support, while Nvidia’s commercial incentives and continuing model work complicate that forecast.
Altimeter and Atreides want Nvidia's board to leave Jensen Huang's return open while it investigates recent events. Their letter backs interim leadership and acknowledges that directors can ultimately reject his reinstatement.
Elon Musk says Nvidia risks losing supply arrangements made personally by former CEO Jensen Huang. No affected supplier is named. Nvidia’s filing documents both $95.2 billion in commitments and limits to the protection those arrangements provide.
The former Anthropic and OpenAI researcher says a pretraining-efficiency slide helped drive his resignation, turning an investor selling point into an argument for a development pause.
Anthropic’s IPO has reportedly attracted orders for six times the available shares at proposed terms targeting a roughly $1.4 trillion valuation. The sale would add to substantial existing cash reserves as the Claude developer balances costly capacity shortages against long-term computing commitments.
Jason Clinton is urging network isolation as AI accelerates vulnerability discovery. The PHP emergency illustrates the immediate repair problem, while Anthropic's own security work shows that finding flaws does not automatically reduce risk.
Former Nvidia CEO Jensen Huang portrayed his relationship with a customer’s executive as routine business, according to reviewed messages and unnamed sources. A board inquiry faces questions about his disclosures amid separate concerns over chip exports and alleged spying.
Apple has detailed how Secure Exclaves isolate background camera and microphone processing from the main operating system and verify visible indicators before releasing full sensor data. The design aims to keep raw ambient sensing data beyond the main operating system’s reach, but its formal verification remains incomplete.
Charta Health is expanding its insurer tools with a $55 million Series B. Its promise of billing accuracy across payers and providers faces a harder test: showing that better coding reduces disputes rather than simply changes payments.
Operand has raised $31 million across seed and Series A funding for AI due diligence software aimed at private equity firms. Its pitch is to reconstruct businesses from public records and internet traces, but evidence of accuracy and deal-level economics remains limited.
Caitlin Kalinowski resigned over the deliberation and oversight behind OpenAI’s Pentagon agreement. The company says its safeguards prohibit domestic surveillance and autonomous weapons, leaving a dispute over how those limits were established and will be enforced.
The startup formerly known as QuilrAI pairs enterprise AI agents with Guardian agents and charges by agents under management. Its expansion pitch faces established alternatives and unanswered questions about security performance and cost.
Luminal’s Standard Capital-led Series A backs software for running AI models more efficiently on existing GPUs and alternative chips. Its reported processing volume leaves revenue and comparable customer savings unresolved.
