A South Korean Tax Accountant Exam Candidate

A South Korean tax accountant exam candidate who placed first with a score of 397.5 out of 400 has admitted to cheating by photographing test questions with a micro-camera and retrieving answers with AI, and the government has moved to void his result and ban him from national exams for five years — exposing how generative AI is undermining professional credential testing, according to Thescoop.

According to reporting by The Scoop, the candidate, identified as Han under the blog nickname “Tax Cat,” sat the first round of the 63rd certified tax accountant exam in April 2026 and the second round in July of the same year, then topped the entire cohort.

His score sheet showed he missed just one of the 160 questions and earned a perfect 100 in tax law fundamentals — the subject that three out of every four candidates failed outright that year under the pass-fail cutoff rule.

The disclosure did not come from the exam authority. It came from Han himself.

‘Genius’ top scorer on South Korea’s tax exam admits to cheating with AI — KOREA NOW

When his pass story and score sheet went up on his blog on October 29, online discussion immediately flagged the anomalous numbers. In an interview broadcast on November 14, Han denied any wrongdoing.

The next day, November 15, he reversed course and admitted he had photographed exam questions with an ultra-miniature camera and obtained the answers through an AI system. He also stated he had used the same method in the first round of the certified public accountant exam.

AI Cheating Breach Casts Doubt on Licensed Tax Professionals

  • A national professional-credential exam now has a documented AI-cheating breach, with an auditor-grade review of the last three to five years of passes now under way — casting doubt on every recently licensed tax professional.
  • The equipment is cheap and accessible: procurement through Seoul’s Yongsan and Sindorim electronics markets was confirmed by one outlet’s reporting, and Chinese-language brokers are selling cheating “services” over Telegram for the college entrance exam, civil service exams and professional credentials.
  • Passing the exam is no longer the moat. The durable value of tax accountants, CPAs and lawyers is shifting from answering set questions to making judgment calls on unstructured client situations.

Top Scorer Full Marks Triggered Investigation

The exam statistic alone explains why the case unraveled publicly. On a paper where 75 percent of test-takers failed at least one subject and were disqualified regardless of aggregate performance, the top scorer posted full marks on the admission-killer subject. Numbers that extreme invite scrutiny, and they got it.

The timeline moved fast in November. The government opened proceedings on November 16 to void the exam result and impose a five-year suspension of eligibility to sit for exams. Police questioning began on November 17.

In 2025, generative-AI-driven group cheating disputes erupted across Korean universities. In 2026 alone, AI-assisted smart glasses were caught at the national technical qualification exams and the TOEIC test. Each episode lowered the perceived and practical barrier to the next.

Chinese-Language Brokers Sell Cheating Services via Telegram

Here is the part that should concern the institutions, not just the test-takers. Cheating assistance has taken on the characteristics of a commercial market. Chinese-language brokers advertise equipment and services over Telegram for Korea’s college entrance exam and a range of civil service and professional credential exams, and some vendors now claim — in their own promotional material — that they helped the tax accountant exam’s top scorer.

The hardware requires no specialist procurement. Reporting cited in The Scoop’s article confirmed the relevant miniature cameras and communications gear could be bought through the Yongsan and Sindorim electronics district. Sophistication is going down while detection exposure is going up, a combination that will stress-test exam security design across every licensing body.

The government’s response extends backward in time as well as forward. The government has begun re-examining the scores of passers from the last three to five years of the tax accountant exam. As the article itself cautions, treating unconfirmed candidates as fraudulent by default would be improper — but with the method proven feasible, the same question now attaches to other national credential exams held in recent years.

What the Market Is Actually Pricing

The event’s second-order debate is the one with a pay-per-hour dimension: if AI can score near-perfect on the tax accountant exam, does the credential still signal anything? Skeptics argue that accountants and lawyers face replacement on the same timeline. That concern applies most acutely to knowledge-and-document work, the sector that generative AI touches first, before physical AI arrives.

But the analysis in the source material draws a sharper line than the raw “AI replaces professionals” argument. The exam tests structured problems: select the correct answer in round one, apply tax law to a fixed set of facts in round two. A real client brings no such structure.

Consider a business owner planning to hand the company to a child. Explaining the legal requirements of the family-business gift-and-inheritance deduction is within reach of an AI tool today. Designing the succession plan around the family relationships, the company’s financial position, the shareholding structure, future capital plans and the possibility of a management dispute is judgment work.

The acknowledgment cuts both ways. Routine tax filing and bookkeeping have been automated for years, and AI-assisted services are improving quickly.

If a client can use AI directly to find a suitable resolution, the professional fee loses its justification — the difference is that the client then carries the legal and economic consequences of the decision personally. Until AI can perform those decisions reliably, credibly, and under liability, the professional layer has structural support.

Exam data point Figure Context
Top scorer’s result, 63rd tax accountant exam 397.5 / 400 Missed 1 of 160 questions
Tax law fundamentals score 100 / 100 3 in 4 candidates failed the subject outright
Eligibility sanction 5 years Exam invalidation proceedings opened November 16
Score review window 3–5 years Past passers’ results under re-examination

The Narrowing Entry Ladder

The personnel consequence lands unevenly. Judgment develops through years of handling routine, low-stakes work — exactly the work automating first. If the first rungs disappear, the ladder by which firms produce senior professionals shortens. For junior tax accountants, accountants and lawyers, the window in which to build problem-solving capability rather than test-taking capability is narrowing, not opening.

That squeeze does not stop at the licensed professions. The same logic applies to knowledge and information work generally, which is the front line of generative AI’s labor impact. Even the author of the underlying analysis reports using ChatGPT and Claude in his own work and is considering Obsidian-based knowledge management and Hermes-style AI agents — a working writer repositioning himself under the same pressure.

The candidate’s own dishonesty will be handled through the administrative and police procedures now in motion, with a government decision and investigation formally under way as of November 16 and 17 respectively. His performance numbers — one question wrong out of 160, a perfect subject that knocked out three quarters of the field — stand as the record that started it.

The era in which passing the exam was the finish line has passed, the analysis concludes. A certified professional’s value going forward depends less on how much is known and more on how well a client’s problem gets solved with that knowledge. Every candidate currently preparing for a national credential exam is, in that sense, studying for a test that has already changed.

Photo of author

Newsroom

Newsroom at Archyde.

Cameron Bancroft Claims Entire Australian Team Knew of 2018 Sandpaper Plot