Polymarket and Kalshi are trying to become sources of news and forecasting, not just places where people trade on future events. But as their probabilities spread across major media outlets, new US investigations into suspected insider trading are testing a central promise of prediction markets: that prices can provide useful information about what is likely to happen next.
Prediction markets let users buy contracts tied to real-world outcomes, whether an election result, economic announcement or geopolitical event will occur. As traders put money behind their expectations, the resulting price is often presented as an implied probability.
That data is now entering mainstream journalism. Polymarket has partnerships with Dow Jones and Substack, where writers can embed live market probabilities directly into articles. It also publishes newsletters and produces the podcast What Are the Odds?. Kalshi, meanwhile, supplies prediction data to CNN, CNBC and Fox, where its probabilities can appear alongside traditional polling and reporting. These partnerships primarily integrate data rather than letting audiences trade directly through news outlets.
Unlike a poll, a prediction market continuously updates as traders react to new information. But that feature creates an obvious vulnerability if some participants possess information unavailable to everyone else.
Documents obtained by WIRED show that the US Commodity Futures Trading Commission opened at least three previously undisclosed investigations involving Polymarket trades. One concerns contracts linked to pardons issued at the end of Joe Biden’s presidency, after an anonymous trader reportedly earned more than $300,000 predicting several specific pardons. Another concerns Iran-related contracts, after a group of accounts reportedly made $2.4 million with a 98% win rate. A third investigation examines additional potential insider trading around Google’s 2025 Year in Search rankings. None of those investigations by themselves establish wrongdoing.
The Google case has already produced public charges. The CFTC and US prosecutors accuse former Google engineer Michele Spagnuolo of using confidential information to make approximately $1.2 million trading Year in Search contracts on Polymarket. The allegations remain subject to court proceedings.
This creates a fundamental tension as prediction markets become information products. A market price can aggregate collective knowledge, but it can also move because a small number of well-informed traders know something the public does not. For media companies increasingly displaying those probabilities as newsworthy data, understanding that distinction may become as important as the number itself.