South Korea deploys real-time AI system to detect crypto manipulation

South Korea’s Financial Supervisory Service said Aug. 20 that it has deployed a real-time, generative AI-powered surveillance system built entirely in-house to detect crypto price manipulation, wash trading and coordinated trading schemes.

The move gives AI Crypto surveillance a much broader role in South Korea’s regulatory framework. Instead of waiting for investigators to manually sift through enormous amounts of order and transaction data, the technology can continuously search thousands of digital assets traded around the clock.

The FSS said its system looks for unusual price and volume movements and compares them against manipulation patterns identified through previous investigations. Among the tactics it can identify are so-called “racehorse” trading, where orders are concentrated during particular short periods to push an asset higher, and “cage” manipulation, which exploits temporary restrictions on deposits or withdrawals to influence prices.

The technology builds on an earlier AI function introduced by the regulator in January to identify the order-book and price-impact periods associated with suspected manipulators. The FSS subsequently added tools capable of identifying groups of linked accounts involved in potential manipulation.

AI looks beyond the order book

The most significant feature of the new AI Crypto surveillance framework may be its ability to connect market movements with information outside the trading system.

When a cryptocurrency experiences a sudden price or volume spike, generative AI can examine exchange announcements, news and other available information to determine whether there is an identifiable explanation. A major listing, network development or other legitimate announcement could account for a sharp move.

If no obvious catalyst exists, the alert can be escalated for closer examination.

The system also targets artificial trading activity, including suspected wash trading and coordinated transactions. The FSS says it evaluates whether trading volumes and transaction patterns deviate from normal distributions and combines multiple indicators before determining whether an asset or period deserves further investigation.

That approach is important because abnormal volume alone does not prove manipulation. A genuine market event can produce extreme price movements. The regulator’s objective is therefore to use AI to prioritize suspicious cases rather than automatically label every unusual trade as illegal.

Online activity is also part of the equation. The FSS has developed technology to detect potential unfair trading through online channels, including illegal advance trading, false information and posts designed to encourage coordinated buying.

That creates a wider surveillance net, covering not only what traders do on exchanges but also information that may be used to influence other market participants.

40 cases put pressure on Korean regulators

The push toward AI Crypto surveillance comes after South Korea spent two years strengthening enforcement under its Virtual Asset User Protection Act, which took effect on July 19, 2024.

By July 2026, South Korean financial authorities said they had completed investigations into roughly 40 cases of unfair virtual-asset trading. More than 30 cases were reported or referred to investigative authorities, involving 25 suspected individuals. Authorities estimated average illicit gains at about 1.4 billion won per case.

The cases included both short-term manipulation and more sophisticated schemes involving multiple exchanges. South Korean authorities have also highlighted manipulation involving API-based trading, coordinated orders and large investors operating across domestic and overseas platforms.

The scale of the problem explains why regulators are increasingly looking toward automation.

An FSS statement said the AI-based surveillance framework should allow the regulator to respond “quickly and efficiently” despite limited personnel as unfair trading becomes more sophisticated.

The next phase could push AI Crypto surveillance beyond identifying suspicious trading patterns. The FSS plans to develop capabilities for tracking fund movements between exchanges and following on-chain transactions, although it has not announced a specific launch date for those features.

That would potentially give investigators a clearer picture of how suspicious funds move after a questionable trade, particularly when activity crosses several platforms.

Human investigators remain the final gatekeeper

Despite the rapid expansion of AI Crypto surveillance, South Korea is not handing enforcement decisions entirely to algorithms.

AI-generated findings are intended to support investigators, who review the reports and determine whether a case warrants detailed analysis or a formal investigation. That distinction is critical because automated systems can identify correlations and anomalies without necessarily understanding the full context behind them.

The experience of U.S. financial regulators illustrates why that safeguard matters. A 2025 Government Accountability Office review found that federal financial regulators were already using AI to identify risks, detect potential legal violations and spot unusual patterns. However, regulators generally said AI outputs were considered alongside other information rather than used as an autonomous basis for decisions. SEC officials, for example, said subject-matter experts review trades flagged by AI when assessing possible insider trading.

The same principle could prove important as South Korea expands its system. AI Crypto surveillance can process market data at a scale that human teams cannot match, but the quality of its conclusions depends on the underlying data, model design and investigative judgment applied afterward.

The Korean experiment therefore represents more than a new monitoring tool. It signals a shift toward continuous, data-driven enforcement in digital assets, where regulators increasingly want to detect manipulation while it is developing rather than reconstruct the evidence after the damage has already been done.

If the planned cross-exchange and blockchain-tracking capabilities are successfully added, AI Crypto surveillance could become an even more comprehensive component of South Korea’s crypto enforcement strategy—and a model other regulators may watch closely.