OpenAI confirmed that it fired an unnamed employee after an internal investigation found the person had used confidential company information in external prediction markets, including Polymarket. The company has not identified the employee, described the trades, disclosed any profit, or said that prosecutors filed a criminal insider-trading case.
The episode is therefore a confirmed employment action and alleged confidentiality-policy violation—not a proven criminal conviction. Separate blockchain analysis identified suspicious activity around OpenAI events, but it did not establish that those wallets belonged to OpenAI employees.
What OpenAI confirmed
In an internal message, Fidji Simo, then OpenAI’s CEO of Applications, disclosed that one employee had been terminated after an internal investigation. OpenAI said the employee used confidential OpenAI information in external prediction markets and cited Polymarket as an example.
Spokesperson Kayla Wood said OpenAI policy prohibits using confidential information for personal financial gain, including through prediction markets. WIRED published the account on February 27, 2026. WIRED’s report is the primary public account of the firing.
#1 Best Overall
OpenAI has not publicly named the employee or released the internal findings. WIRED characterized the event as the first confirmed firing by a major technology company over prediction-market trading, a description that should be understood as that outlet’s characterization rather than an exhaustive global survey.
What remains unknown
- The employee’s name, job title, seniority and location.
- Whether the person traded directly, used another person’s account or passed information to someone else.
- The specific contracts, platforms, dates, amounts, wallets and profits involved.
- Whether the trades were on Polymarket, Kalshi or another market.
- What kind of confidential information was involved, such as a product launch, executive decision or another internal event.
- Whether OpenAI referred the matter to the Commodity Futures Trading Commission, law enforcement or another regulator.
- Whether the employee admitted wrongdoing or was accused of leaking information beyond the trading activity.
Those gaps make it impossible to reconstruct the individual case from public information.
What the wallet analysis found—and what it did not prove
Blockchain-analysis group Unusual Whales reported clusters of activity around OpenAI-related events dating to March 2023. Its review flagged 77 positions across 60 pseudonymous wallet addresses. The activity included markets concerning Sora, GPT-5, the ChatGPT Browser and Sam Altman’s employment status.
Rank #2
- Comes with secure packaging
- Easy to read text
- It can be a gift option
| Finding reported by Unusual Whales | Qualification |
|---|---|
| 77 positions across 60 wallets | Behavioral flags, not proof that the wallets belonged to OpenAI employees. |
| More than $16,000 profit on a November 2023 Altman-return trade | Reported profit for one wallet in the broader analysis; no evidence ties it to the fired employee. |
| 13 newly created wallets bet $309,486 in the 40 hours before the browser launch | Suspicious timing identified by the analysis; wallet ownership, information source and legality remain unestablished. |
Polymarket uses blockchain infrastructure on Polygon. Its trading ledger is pseudonymous but traceable: observers can see transactions and timing, while the real-world identity behind a wallet is not automatically visible. Polymarket did not respond to WIRED’s requests for comment about the suspected activity.
Why suspicious does not mean proven
Investigators may examine newly created wallets, unusually large bets, several accounts taking the same position, trades immediately before an announcement, or an account going inactive after a successful trade. Each can have lawful explanations, including public rumors, copied trades, independent forecasting, coordinated but legal speculation, manipulation, wash trading or attribution errors.
There is a critical evidence ladder:
- Public wallet activity: transactions visible on the blockchain.
- Wallet attribution: evidence connecting an address to a particular person.
- Information provenance: evidence that the person possessed confidential information.
- Intent: evidence that the information drove the trade.
- Legal proof: evidence meeting a regulator’s or court’s standard.
The Unusual Whales figures primarily describe the first level. They do not, by themselves, establish the other four.
Is this legally “insider trading”?
That label remains legally unresolved. Traditional insider-trading cases generally involve securities or other regulated financial instruments, material nonpublic information and a breach of a duty. Prediction-market contracts concern future events rather than conventional company shares, and the applicable rules can depend on the platform, contract, jurisdiction and facts.
OpenAI could fire someone for violating confidentiality, ethics, conflicts-of-interest or personal-gain rules even if no prosecutor brings a case. Conversely, a termination does not itself prove that a crime occurred. The careful description is: OpenAI treated the conduct as a violation of its policy against using confidential information for personal gain; whether it also violated a specific insider-trading law has not been established publicly.
Why prediction markets create an information-integrity problem
Contracts increasingly cover technology launches, corporate decisions, executive departures, earnings and other news-sensitive outcomes. Employees can know about those events before the public, while binary contracts can make a small timing advantage financially valuable. Blockchain records expose when a bet was placed, but pseudonymity makes it difficult to determine who traded and why.
Rank #4
Market wording adds another complication. A confidential milestone may not map neatly to the contract’s resolution rules:
- Does a product preview count as a launch?
- Which announcement date controls?
- Can a delayed or canceled release resolve the contract differently?
- Who decides whether the wording has been satisfied?
These questions affect both fairness and enforcement. A trader might know an internal milestone while still misunderstanding how the market will resolve.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How this compares with other cases
WIRED reported that Kalshi has referred several suspected insider-trading cases to the CFTC and has announced sanctions in other matters. Reported examples include a MrBeast employee suspended and fined $20,000 over trades related to the creator’s activities and a political candidate banned for trading on his own campaign.
Best Value
WIRED also reported that Google, Meta and Nvidia did not respond when asked whether they monitor employees for prediction-market insider trading or maintain relevant policies. A nonresponse is not evidence that those companies lack controls or permit the practice. Separately, WIRED has described government efforts to restrict public employees’ use of nonpublic information on prediction markets, including New York’s ban. That report provides the broader policy context.
What could happen next
Further wallet-attribution work could test whether any suspicious addresses connect to employees or information leaks, but attribution would still need evidence of knowledge and intent. Employers may tighten disclosure, trading and confidentiality rules; exchanges may expand monitoring and referrals; and regulators may clarify how existing laws apply to event contracts.
Unless OpenAI releases more information or authorities announce an action, the public record remains limited to one confirmed firing, an undisclosed internal case and separate analytical findings that have not been tied to the terminated employee.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →
