South Korea watchdog deploys AI crypto surveillance

TLDR

South Korea's main financial watchdog has launched an AI-powered system to detect crypto market manipulation across local and major global exchanges in real time.

1. The Financial Supervisory Service (FSS) is rolling out the ORBIT and VISTA platforms to monitor trading and cluster suspicious accounts across eight exchanges.
2. This raises the risk of enforcement for pump-and-dump schemes, wash trading, and cross-exchange manipulation, but could improve market integrity for Korean crypto users.
3. Next, the system is set to expand into on-chain tracing and text analysis, while other regulators like the CFTC also adopt AI, pointing to a global shift in supervision.

Deep Dive

1. How The New AI System Works

The FSS has introduced an AI surveillance platform called ORBIT (Oversight & Reporting system of Blockchain Irregular Transactions), which ingests minute-by-minute data from five domestic exchanges (Upbit, Bithumb, Coinone, Korbit, Gopax) and three global venues (Binance, Coinbase, OKX), covering about 1,942 tokens as of March. The system flags anomalies in price and volume and can highlight specific tokens for immediate review based on pre-set patterns learned from past enforcement cases such as "racehorse" style pumps and coordinated dumps.

A companion engine, VISTA (Virtual Assets Intelligence System for Trading Analysis), uses machine learning techniques like UMAP and HDBSCAN clustering to group accounts that appear to trade in coordination, helping investigators identify suspected manipulation networks across multiple accounts and exchanges. In early 2026 the bureau has already launched more AI-assisted investigations than it did in the whole of 2025, according to the FSS-focused report on the new AI-powered surveillance system.

2. Impact On Traders And Exchanges

For traders, the biggest change is that previously hard-to-detect behaviors are more likely to be flagged automatically, including rapid cross-exchange arbitrage that looks like wash trading, orchestrated pump groups on illiquid tokens, and bot-driven spoofing. For exchanges, the system increases pressure to align internal surveillance with the FSS view, since ORBIT pulls their data directly and can surface discrepancies between their own monitoring and the regulator's.

This fits a broader Korean pattern: the National Tax Service is separately procuring a "Virtual Asset Integrated Analysis System" that uses AI to track crypto for tax enforcement, as described in its crypto tax enforcement bid. Together, these tools mean both trading conduct and tax behavior in Korean-linked flows could face much tighter scrutiny.

3. What To Watch Next

The FSS roadmap includes a third phase that adds on-chain tracing, LLM-based analysis of text data, and automated document generation, extending surveillance beyond exchange order books into wallets and public communications. That would make it easier to connect off-exchange wallets, social coordination (for example, pump groups), and on-exchange executions into a single case file.

Globally, the U.S. Commodity Futures Trading Commission is also deploying AI to review crypto registration applications and monitor trading, according to its chair's comments on AI-driven oversight. This suggests AI monitoring will increasingly be a baseline expectation in major regulated markets.

Conclusion

South Korea's AI crypto surveillance rollout signals a move from manual case-by-case investigations toward continuous, data-driven oversight of exchanges and trading behavior. That can reduce room for blatant manipulation and tax evasion, but it also raises the stakes for compliance systems and increases the chance that aggressive but legal strategies get scrutinized. For participants active in or routing through Korea, the practical shift is toward cleaner, more documented activity, with less tolerance for opaque structures or coordinated trading games.