Korean Police Traced Polymarket Users Through the Blockchain

The Nature of Predictions Remains Contentious

The Gangwon Provincial Police Agency’s cyber investigation unit opened a preliminary inquiry in March and began booking suspects in May. Korea’s broadcasting and communications standards body did not order domestic access to Polymarket blocked until Aug. 18. Nobody in this case is accused of circumventing that block: the conduct under investigation took place while the platform was openly reachable. Instead, they are straightforwardly charged with gambling.

As of Sept. 15, police had booked 26 people and referred 18 to prosecutors, according to National Police Agency materials first reported by Asia Business Daily. Those booked staked a combined $12.7 million (KRW 17.6 billion), with the largest single trader accounting for $4.1 million (KRW 5.7 billion). The investigation is continuing, and Kangwon Ilbo reported that both the number referred and the transaction totals may grow.

The way police tracked down its suspects has consequences beyond Korea. Polymarket runs as a non-custodial peer-to-peer market with automatic settlement, and held no real-name roster of Korean users. Police said obtaining a user list from the operator was therefore unrealistic, but that blockchain transaction records are public, so they could identify and track individual users through open-source intelligence analysis of the chain itself, as Digital Asset reported.

The charge is illegal gambling under Article 246 of Korea’s Criminal Act. Supreme Court precedent holds that gambling can be established even where a participant’s own skill influences the result, provided chance also intervenes to any degree. Police did not identify the case, but that formulation matches a 2008 judgment in which four men were convicted of habitual gambling for playing golf for money, with the court finding that however much skill golf demands, no player can predict a hole with certainty or control where the ball comes to rest. Resemblance to derivatives investment, police said, and the absence of specific regulatory guidance do not by themselves exclude the gambling provision.

This is the main argument across the globe when it comes to deciding what a prediction market is, and how its products should be classified. The pro-market argument is that users analyze political, economic, and social information to price the likelihood of an event, trade contracts on an order book, and can sell a position before resolution to lock in a profit or loss, contending this makes it a crypto-based derivatives market rather than a bet. It is a version of the same classification fight running through Europe, where ESMA concluded the EU retail ban already covers many prediction markets.

Kim Tae-rim, an attorney at AXIS Law, told Asia Business Daily the formal elements of the offence may well be satisfied (crypto was staked, a chance-element was present) and that Korea’s personality principle makes domestic criminal law difficult to escape. The decisive question, he said, is how a court assesses the structural difference of order-book contracts that can be liquidated before expiry.

When the block was ordered in August by the national broadcasting and communications standards body, the standards body cited a winner-take-all payoff structure on events users cannot control. Polymarket responded that it offers no Korean-language service, accepts no payments made in won, and cannot be treated as the operator of a gambling venue because it never holds user funds, but these arguments were rejected by the authorities.

Polymarket has been blocked or pursued in a widening list of jurisdictions, primarily in Europe: Czechia ordered a 15-day ISP shutdown in July, Romania’s ban held up in court in April, and the Dutch regulator imposed a €487,000 penalty in June. Korea’s approach of targeting players is novel but isn’t without precedent: Tokyo police referred ten customers of overseas online casinos who gambled with crypto in November 2024, part of a sweep in which 24 prefectural forces caught 57 of roughly 130 identified players. The stakes and the product here make it a very different proposition.