Can a regulator catch market manipulators when the markets themselves sit beyond its jurisdiction? The US Commodity Futures Trading Commission is betting that artificial intelligence can bridge that gap—and in doing so, fundamentally reshaping how financial enforcement works.
The CFTC, historically one of Washington's leanest agencies, has begun deploying AI surveillance tools to identify US-based traders accessing offshore prediction markets like Polymarket. Chairman Michael Selig delivered a blunt warning from the agency's Washington headquarters: "We're going to find them, and we're going to bring actions." The targets: Americans who use commercial VPNs—services costing roughly $10 monthly—to circumvent geographic blocks on platforms legally operating beyond US borders.
Polymarket, registered in Dublin with operations in Seychelles, holds no US license and officially bars American IP addresses. Yet the platform has processed billions in bets on geopolitical events. Some traders made fortunes on suspiciously timed positions around the Venezuela raid and the Iran conflict. Whether those positions reflected inside information or market manipulation remains under investigation—but the CFTC now has automated tools designed to catch such patterns.
"You've got so much data," Selig told WIRED, describing the agency's new AI capabilities. "When we feed it into AI, we get really great information. It can help us understand things, like where we might want to investigate, or when we might need to send a subpoena to a trader."
The implications extend beyond prediction markets. The CFTC's approach establishes what may become the dominant regulatory model for the AI era: AI monitors, flags suspicious activity, and human investigators decide whether to pursue cases. This "machine surveillance, human judgment" framework offers clear advantages—AI systems can process trading volumes no human team could analyze. They detect subtle patterns across millions of transactions. They never tire or face budget constraints.
But the approach also surfaces genuine concerns. VPNs deliberately obscure user identity through layers of intermediaries. The CFTC has no jurisdiction over offshore platforms and cannot compel Polymarket to hand over customer records without international cooperation that currently doesn't exist. Selig acknowledges these constraints but frames AI surveillance as the necessary first step: you cannot enforce what you cannot detect.
The deeper tension lies in whether technological surveillance alone can close an enforcement gap rooted in legal architecture rather than detection capability. Americans face potential liability for trades placed on platforms the CFTC cannot directly regulate. The agency may flag their positions, build cases against their activity—but prosecution still depends on evidence gathered across jurisdictions that have no obligation to assist.
For now, the CFTC's AI surveillance represents a significant shift in regulatory posture—one likely to accelerate across financial watchdogs. Whether it can penetrate the offshore economy remains genuinely uncertain. But the infrastructure being built today suggests the enforcement gap will eventually narrow. Traders operating in gray zones should assume they are being watched.