OpenAI Fires Employee for Insider Trading on Polymarket Prediction Markets - Explained

OpenAI has fired an employee after an internal investigation concluded they traded on prediction market platforms using nonpublic company information, according to WIRED’s reporting. The firing is significant because it’s a rare, explicit corporate enforcement action tied to event-contract trading rather than traditional securities trading.

The core allegation is straightforward: an employee with access to confidential information used that informational edge to place trades on event contracts tied to OpenAI outcomes. On prediction markets, those contracts typically pay out $1 if a specified event happens (and $0 if it doesn’t), meaning any trader who can reliably anticipate the outcome before the rest of the market can buy the right side early and capture value as prices move. The employee’s activity involved multiple prediction markets - including Polymarket.

What’s important from a compliance perspective is what OpenAI is treating as the breach. This is not about securities trading or public-company earnings leaks in the traditional sense. It is about the misuse of confidential corporate information for personal gain in a different, fast-growing market structure - one where the underlying subject matter is often operational decisions that a subset of employees may know before anyone else.

Fidji Simo, OpenAI’s CEO of Applications, disclosed the termination in an internal message to employees earlier this year, writing that the employee “used confidential OpenAI information in connection with external prediction markets (e.g. Polymarket).”

That internal framing matters because it indicates OpenAI is approaching prediction-market trading the same way many companies approach restricted securities trading: if you possess material, nonpublic information about the business, you cannot use it to place a trade that benefits you personally - regardless of whether the venue is a stock exchange or a prediction market. 

OpenAI spokesperson Kayla Wood gave a statement that makes that policy boundary explicit: 

“Our policies prohibit employees from using confidential OpenAI information for personal gain, including in prediction markets.”

What OpenAI has not put on the record is equally notable. The company has not identified the employee, described the specific markets they traded, disclosed the size of the positions, or said whether it referred the matter to regulators or law enforcement.

So, as of now, the public record supports a clear finding but leaves open several operational questions readers will care about: when the trades occurred, what internal information was used, and what steps - if any - OpenAI is taking beyond the employment action to prevent this from happening again.

OpenAI’s firing is unlikely to be the last corporate action of this kind, because prediction markets now cover exactly the kinds of things employees can learn early: launches, leadership changes, partnerships, and regulatory decisions.

Three developments to watch:

• Corporate controls: More explicit employee policies around event contracts (not just traditional securities) and tighter disclosure/approval rules for outside trading.
• Platform enforcement: Kalshi is building a record of investigations, penalties, and referrals, and has described surveillance triggers like statistically anomalous success and user tips.
• Regulatory posture: The CFTC is putting a finer point on what it considers actionable misuse of nonpublic information on regulated venues - useful guidance for platforms and traders, and a potential template for future cases.

For bettors, the practical takeaway is simple: treat markets tied to corporate or institutional decision-making as higher risk for information asymmetry, and expect more rules tightening as companies and regulators react.