News

OpenAI Fires Employee for Prediction Market Insider Trading - And the Data Suggests They Were Not Alone

OpenAI terminated an employee for using confidential company information to trade on Polymarket, the first confirmed firing of its kind at a major AI lab. An Unusual Whales analysis of on-chain data found 60 suspicious wallets and 77 positions tied to unreleased OpenAI products.

OpenAI Fires Employee for Prediction Market Insider Trading - And the Data Suggests They Were Not Alone

Forty hours before OpenAI publicly launched the ChatGPT Browser, thirteen brand-new cryptocurrency wallets - each with zero trading history - appeared on Polymarket for the first time. Collectively, they wagered $309,486 on the correct outcome. Every single one of them won.

Two days after Sam Altman was fired as CEO in November 2023, a fresh wallet materialized on the same platform, placed a single bet predicting his return, collected over $16,000 in profit, and never traded again.

These aren't hypotheticals. They're on-chain records, timestamped and immutable, sitting on a public blockchain for anyone to verify. And on Thursday, OpenAI confirmed what the data already suggested: at least one of its employees was using inside information to trade on prediction markets. The company fired them.

What OpenAI Found

OpenAI's CEO of Applications, Fidji Simo, disclosed the termination in an internal message to employees. The company's spokesperson, Kayla Wood, confirmed the firing publicly: "Our policies prohibit employees from using confidential OpenAI information for personal gain, including in prediction markets."

The unnamed employee used confidential information about unreleased OpenAI products to place bets on Polymarket - the largest prediction market platform, where contracts on everything from elections to product launches trade on a public blockchain. OpenAI didn't disclose how it discovered the trading, what products were involved, or how much the employee profited.

What OpenAI didn't say is more interesting than what it did.

The Unusual Whales report

Financial data platform Unusual Whales conducted an independent analysis of Polymarket trading activity around major OpenAI announcements dating back to March 2023. Their findings are stark.

MetricValue
Suspicious wallets identified60
Flagged positions77
ChatGPT Browser pre-launch bets$309,486 across 13 new wallets
Time window before launch40 hours
Altman return bet profit$16,000+ from a single wallet
Events with suspicious patternsSora, GPT-5, ChatGPT Browser

"When you see that many fresh wallets making the same bet at the same time, it raises a real question about whether the secret is getting out," said Unusual Whales CEO Matt Saincome.

OpenAI fired one employee. The data suggests the problem is far larger than one person.

Trading charts on multiple screens showing market activity Prediction markets now generate $60 billion in annual volume - and the infrastructure to police insider trading hasn't kept pace.

A Pattern Across the Industry

OpenAI's firing is the first confirmed case of a major tech company terminating someone over prediction market trading. But it arrived during a week when the entire prediction market ecosystem began reckoning with the same problem.

Kalshi vs. the MrBeast editor - On February 25, regulated prediction market Kalshi announced it had fined Artem Kaptur, a video editor for YouTube star MrBeast, $20,397 for using insider knowledge about upcoming MrBeast videos to trade on the platform. Kaptur wagered roughly $4,000 on YouTube streaming markets in August and September 2025, earning $5,397 in profits. Kalshi also banned him for two years and referred the case to the CFTC.

The California politician - In the same enforcement batch, Kalshi suspended Kyle Langford for five years and fined him $2,246 for betting roughly $200 on his own candidacy for Governor of California, then posting about it on social media.

Israel's ricosuave666 - In the most extreme case, Israel brought the first criminal prosecution in history for prediction market insider trading in February 2026. A trader using the handle "ricosuave666" placed seven bets on Israel-Iran strike timing during the June 2025 conflict, reaching a perfect prediction record on classified operational questions. The trader earned approximately $150,000.

The AlphaRacoon affair - A wallet deposited $3 million on Polymarket in December 2025, made 22 of 23 correct predictions on Google "Year in Search" markets, and netted $1.15 million in under 24 hours. The same wallet had previously profited over $150,000 predicting Gemini 3.0's launch window. No enforcement action has been taken.

The $60 Billion Gap in the Law

Prediction markets are now a $60 billion annual industry - up roughly 400 percent from 2024. Bloomberg terminals carry their data. CNN integrates Kalshi markets into live coverage. Wall Street institutions including DRW, Susquehanna, Jump Trading, and Goldman Sachs have established dedicated trading desks. The Federal Reserve's economists published research in February 2026 treating Kalshi's macro markets as "credible information infrastructure," noting they beat professional forecasting surveys on FOMC meeting predictions.

The problem: the legal framework governing these markets wasn't built for this.

Prediction market contracts are structured as swaps, placing them under CFTC jurisdiction rather than SEC oversight. The CFTC's anti-fraud rule, Rule 180.1, is modeled on the SEC's Rule 10b-5 but with a critical distinction - it requires proof of a "breached pre-existing duty." In securities law, nearly any trade based on material nonpublic information violates regulations. In commodities law, proprietary information trading is often the market's intended mechanism.

The result: the CFTC has started zero enforcement actions specifically targeting prediction market insider trading despite at least three major documented cases since December 2025.

SDNY U.S. Attorney Jay Clayton signaled a harder line at the Securities Enforcement Forum on February 5. "Because it's a prediction market doesn't insulate you from fraud," he said, adding that enforcement actions should be "expected." But wire fraud prosecutions require proving the platform's terms of service were violated - and Polymarket, operating through a Panama-incorporated entity with no identity verification requirements, doesn't explicitly prohibit insider trading in its terms.

A judge's gavel on a legal desk SDNY U.S. Attorney Jay Clayton warned in February that prediction market participants are "not beyond the reach of fraud statutes."

The structural paradox

Economist Eric Zitzewitz has pointed out the fundamental tension: prediction markets "require loads of uninformed investors to function." When insiders consistently profit at extreme win rates, retail participants recognize they are being methodically picked off and withdraw. This triggers what economist George Akerlof described as the "market for lemons" dynamic - as informed money drives out uninformed money, liquidity collapses, and the market's ability to aggregate information dies.

"The problem isn't that you have information the market doesn't. The problem is that you're using information that belongs to someone else - your employer or client or country - without their permission."

  • Matt Levine, Bloomberg Opinion

The irony is structural: insider trading simultaneously improves short-term price accuracy while destroying long-term market viability. The very thing that makes prediction markets useful - their ability to aggregate information from people who know things - is also the thing that'll kill them if left unchecked.

What OpenAI Is Not Saying

Meta, Nvidia, and Google didn't respond to requests about whether they monitor employee activity on prediction markets or maintain similar policies. OpenAI's policy explicitly covers prediction markets, but it's unclear how many other AI companies with billion-dollar valuations have equivalent guardrails.

The broader question is whether a single firing addresses what appears to be a systemic problem. Sixty wallets with 77 suspicious positions across multiple OpenAI product launches doesn't describe a lone actor. It describes a culture - or at the very least, a widely shared understanding that the rules hadn't yet been written.

OpenAI has built one of the most valuable technology companies in history, with 900 million weekly active users and $20 billion in annual recurring revenue. Every product decision it makes moves markets. The company's launch calendar is now, quite literally, a financial instrument.

What Comes Next

The prediction market insider trading crisis is converging on three fronts simultaneously:

  1. Corporate enforcement is reactive and inconsistent. OpenAI fired one employee; the data suggests dozens were trading. Meta, Google, and Nvidia haven't disclosed whether they even have policies covering prediction markets.

  2. Platform enforcement is split. Kalshi runs an internal surveillance system called "Poirot" and has completed over 200 investigations annually. Polymarket operates through a Panama entity with no KYC, no explicit insider trading prohibition, and has not publicly addressed the OpenAI wallet cluster.

  3. Government enforcement is nascent. SDNY has signaled intent. Israel has brought the first criminal prosecution globally. The CFTC has announced rulemaking plans but zero actual enforcement. Rep. Ritchie Torres has introduced legislation banning federal officials from trading prediction markets related to government activity, but it doesn't cover corporate insiders.

Matt Levine's prediction may be the most honest assessment of where this ends: the first twenty prediction market insider trading arrests "will be very surprised."

OpenAI's unnamed employee won't be the last person fired over this. The question is whether they'll be among the first prosecuted.

Sources:

OpenAI Fires Employee for Prediction Market Insider Trading - And the Data Suggests They Were Not Alone
About the author Senior AI Editor & Investigative Journalist

Elena is a technology journalist with over eight years of experience covering artificial intelligence, machine learning, and the startup ecosystem.