The CFTC Is Done Watching from the Sidelines

CFTC Chairman Michael Selig made it plain this week: the agency is coming for insider traders in prediction markets, and it's using AI to do it.

"We're going to find them, and we're going to bring actions," Selig told WIRED directly, speaking from CFTC headquarters in Washington, D.C.

This marks a significant escalation. For most of the past year, the offshore, crypto-based nature of platforms like Polymarket left serious questions about whether U.S. regulators would act. This week answered them.

The Tools on the Table

The CFTC isn't working with hunches. According to WIRED, the agency is running a multi-layered surveillance apparatus that includes proprietary in-house trading pattern analysis, Chainalysis for blockchain tracing on crypto platforms, and Nasdaq Smarts for abuse detection on centralized markets.

Selig explained the volume problem bluntly: "You've got so much data. 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 agency admits it's lean on staff right now. It's staffing up — and using automation to bridge the gap.

The Numbers That Make This Urgent

This isn't a niche problem. Prediction markets processed more than $44 billion in wagers last year, according to reporting via Cryptopolitan and MEXC News.

Polymarket alone registered $10.2 billion in bets in April — down from $11.2 billion in March, an 8.9% drop and its first monthly decline since August. Rival Kalshi, by contrast, saw volume jump 13% to $14.8 billion in April. Dune Analytics is the source on those figures.

Polymarket is losing ground at exactly the moment it faces the most scrutiny.

The Bot Problem Nobody Is Talking About Enough

Most mainstream coverage glosses over a critical dimension: the insider trading problem is getting harder to solve because bots are now running the show.

On Polymarket, automated bots now control more than 30% of active accounts, according to Cryptopolitan. Of the top 20 earners on the platform, 14 are bots. More than 37% of automated accounts consistently outperform human traders.

Regulators face a two-front problem. Humans with classified intel making illegal bets. And bots potentially being weaponized to launder the profits or mask the trading patterns. Regulators are fighting this with AI — but so are the bad actors.

What Congress Is Doing About It

Senator Elizabeth Warren, along with more than 40 other members of Congress, sent a letter to the CFTC in March demanding laws that would ban government officials from profiting off classified material on prediction markets, according to MEXC News.

The lawmakers argued that the CFTC should "ensure that federal employees understand existing restrictions on prediction market insider trading."

It's a reasonable demand. But it comes late. The Army sergeant indictment came after the damage was already done. Rules that weren't enforced are barely rules at all.

The Hard Data on Defense Bets

This goes beyond one soldier. The Anti-Corruption Data Collective analyzed more than 400,000 prediction markets settled on Polymarket over five years, according to the New York Times' Hard Fork podcast.

Result: long-shot bets tied to military or defense events had an average win rate of roughly 52%.

The platform-wide average win rate is 14%.

Military bets win at nearly four times the baseline rate. Either a lot of people are incredibly lucky about classified operations, or classified information is flowing into these markets on a regular basis.

Polymarket's Scramble to Clean Up

Polymarket has spent this spring trying to rebuild credibility. In April, the company announced a partnership with Chainalysis for its offshore platform. It also updated its market integrity rules and announced a partnership with Palantir for its U.S.-based sports markets, according to Ars Technica and WIRED.

Notably, Polymarket CEO Shayne Coplan previously argued that insider trading could actually be good for prediction markets — a position the company has now abandoned. The company did not respond to WIRED's requests for comment.

Kalshi, meanwhile, is playing offense. The U.S.-based exchange publicly announced it has suspended and penalized customers flagged for insider trading and market manipulation. That shift in strategy appears to be working: Kalshi's volume is surging while Polymarket's slips.

What Mainstream Coverage Is Missing

Left-leaning outlets are framing this primarily as a regulation story — Wild West crypto markets finally getting reined in by government oversight. That's one angle.

But the real story has two layers most coverage ignores. First, the bot dominance problem means the insider trading crackdown is chasing one symptom while a structural infection spreads. Second, the CFTC being lean on resources right now raises a hard question: does this agency have the capacity to actually police a $44 billion market with AI tools and a skeleton crew?

Sending subpoenas is not the same as winning prosecutions.

Where This Stands

Regulators are finally moving. AI surveillance is live. Polymarket is hemorrhaging market share to a cleaner competitor. Congress is making noise.

But the CFTC is a lean agency trying to police a massive, fast-moving, bot-saturated market where classified government intel has been flowing in for over a year. The Army sergeant case was the first criminal charge — and by the numbers, almost certainly not the last.

Regular people using these platforms for legitimate purposes are now betting alongside bots and insiders in a market where the edge often belongs to whoever has access to information the public isn't allowed to know. A few Chainalysis contracts and AI flags won't fix that overnight.