South Korea’s Financial Supervisory Service has developed an artificial intelligence-based system to monitor the digital-asset market for signs of unfair trading in real time. The platform is designed to detect conduct such as price manipulation, wash trading and coordinated activity across linked accounts.
According to the source report, the system combines generative AI with machine-learning models and draws on a broad set of inputs, including exchange data, notices, news reports, online chat content, videos and social-media-style posts. Its purpose is to identify suspicious patterns more quickly and help officials decide whether a formal investigation should begin.
Real-time monitoring of market abuse
The FSS said the surveillance system watches for ultra-short-term pump-and-dump activity as it happens. That includes patterns described in the report as racehorse and pen types, as well as abrupt and unusual moves in token prices and trading volumes.
In addition to spotting sudden market anomalies, the system is built to identify accounts that may be acting together in manipulative trading. By screening live exchange data, it flags activity that appears abnormal and then uses AI to review related notices and news as part of the assessment process.
How AI supports case selection and analysis
The agency’s use of AI goes beyond initial alerts. After unusual trading is detected, the technology helps with deeper analysis by selecting tokens for review based on real-time surveillance results, investor complaints and media reports.
The report said AI also assists investigators by drafting review documents. That means the system is being used not only as a detection tool but also as support infrastructure for deciding which cases merit closer scrutiny and possible enforcement action.
Wash trades, front-running and online promotion
For wash trading, the FSS said it applies Benford’s law together with machine-learning and anomaly-detection techniques. The goal is to identify trading patterns that may indicate fake or self-matched transactions intended to create misleading market activity.
The surveillance framework also extends beyond order-book and transaction data. It examines possible illegal front-running in online chat rooms, checks videos for false information and reviews posts that may encourage unfair trading practices. To do that, the system uses APIs to gather content, convert video audio into text and assess the likelihood that material violates market rules.
Next planned upgrades
The FSS plans to expand the system further by adding fund-flow analysis and on-chain tracking. Those additions would widen the scope of its monitoring from exchange-based trading behavior to the movement of assets as well.
The regulator’s stated aim is to respond faster and more efficiently to suspected misconduct while strengthening its broader AI-based surveillance and investigative framework. In the source report, that effort was framed as part of protecting users and maintaining orderly conditions in the digital-asset market.
Source: en.bloomingbit.io