Military accounts flagged by Polymarket surveillance tools

Polymarket surveillance tools flagged military accounts before public report

In April, a grand jury in the Eastern District of Virginia indicted Army soldier Gannon Ken Van Dyke on charges including commodities fraud and theft of nonpublic government information. The allegation was that he had used classified military intelligence to place profitable wagers on Polymarket. The indictment made the platform's name visible in a federal courthouse, which is not where any prediction market operator wants to be.

What CNN has now reported changes the sequence in a way that matters: Polymarket referred dozens of accounts showing potential insider activity to the Department of Justice before outside researchers published their findings. The Anti-Corruption Data Collective later identified 152 accounts that earned roughly eight million dollars across war-related markets. Polymarket had already flagged the problem internally and sent it upstream.

The platform's surveillance system monitors approximately 150 distinct signals and trading patterns. The company has not disclosed what those signals are, which is the correct decision — publishing your detection criteria is an instruction manual for evasion. What it has disclosed, through the CNN report, is that it was not waiting for academic researchers or federal investigators to tell it something was wrong.

The conventional read is that this is reputation management. A company under CFTC investigation and congressional scrutiny releases information showing it acted first. That reading is probably right, and it does not make the underlying fact any less significant. The sequence matters independently of why it was disclosed.

Here is where the consensus framing misses something. Most of the coverage has treated this as a question of Polymarket's integrity posture — did the company do enough, was it fast enough, is it cooperating sufficiently. That is the wrong frame for what the market is actually pricing. The harder structural problem is that referrals to the DOJ after suspicious activity occurs are not the same thing as preventing a trade from being placed on nonpublic information in the first place. The two things are categorically different.

Kalshi has built its integrity infrastructure around the second problem. Employment verification for traders in high-risk markets, national security risk as a factor in contract evaluation, around-the-clock account monitoring — these are ex ante controls. Polymarket's disclosed approach is predominantly ex post: flag, refer, freeze. In a previous position, the distinction between pre-event controls and post-event reporting was the difference between a finding that led somewhere and one that didn't. Regulators know this. Congressional staff know this. The CFTC knows it well.

Whether Polymarket has ex ante controls it has not disclosed is not on the public record. What is on the record is that the Anti-Corruption Data Collective found the accounts through open trading data, and Polymarket found them through internal surveillance. Both methods identified the problem after the money had moved.