A Google Engineer Made $1.2M Insider Trading on Polymarket

Michelle Spagneulo, a 36-year-old Italian citizen living in Switzerland and employed as a Google engineer, made approximately $1.2 million trading on Polymarket under the username alpha raccoon. He used confidential Google internal data to bet that a person named David would be Google's most searched person of the year, placing bets between October and December when the prediction market assigned near-zero probability to that outcome. David's implied probability rose from 0.1 percent on November 28 to around 10 percent just before market resolution. Spagneulo is charged in the Southern District of New York with one count each of commodities fraud, wire fraud, and money laundering, and prosecutors allege he took deliberate steps to conceal the source and ownership of his proceeds.

A separate Polymarket user with the username betwick, apparently not the insider, independently turned $3,000 into $125,000 by betting on David at a 2 percent likelihood. Someone on the other side of Spagneulo's trades sold tokens cheaply without knowing the counterparty held inside information, a dynamic the speakers acknowledge feels unfair even if insider trading is sometimes characterized as a victimless crime. Jay Clayton is quoted stating that corporate insiders cannot use confidential business information to turn a profit in markets and that insider trading compromises market integrity.

This is the second significant criminal case tied to Polymarket trades. A US soldier was charged in April for betting on a US operation to capture Nicolas Maduro and made more than $400,000 in profit. In June 2025, Israeli authorities charged a reservist and a civilian for using classified military information to profit from Polymarket bets tied to Israeli strikes against Iran, with approximately $150,000 in profit involved. A CBS 60 Minutes segment working with Bubblemaps separately identified accounts with a 98 percent win rate on US military action markets on Polymarket. Insider trading incidents have also occurred on Kalshi, and the speakers predict the Google case will not be the last such instance on prediction markets generally.

Smaller-scale cases have emerged on Kalshi as well. A Mr. Beast editor was caught trading on Kalshi using knowledge of a video outcome and made $5,000 in profit. Former California gubernatorial candidate Carl Langford was banned from Kalshi for five years and fined $2,000 after betting on his own race and promoting it. Byron Gilliam argued that prediction markets can only provide better information to the world if insiders are free to participate, framing a genuine tension between market integrity enforcement and the informational value that insider participation theoretically provides.

Manuel Arios, co-founder and former CTO of Open Zeppelin who left the company around 2019, has been privately advising friends and family to exit all DeFi positions, including blue-chip protocols such as Aave, MakerDAO, and Compound. His argument is that coding agents are now superhuman at finding smart contract vulnerabilities and that the security asymmetry is too severe for defenders, with his concern extending beyond smart contract code to parameter configuration, mechanism design, and operational security. Open Zeppelin publicly clarified that his posts do not reflect the company's position and that AI is also one of the most powerful defensive tools available when used with rigor and expert human judgment.

Mark Zellow pushed back, arguing that less than 10 percent of DeFi issues over the past year stem from the code base itself, with most failures attributable to bad parameter configuration, collateral blowups, and poor operational security rather than exploited smart contract code. Speakers acknowledged that attackers currently appear to hold the advantage but suggested the pendulum could swing as AI-assisted defense matures. The broader concern raised is that if DeFi yields only one or two percentage points above fiat bank accounts, those returns may not justify the risks of self-custody and smart contract exposure, and that if DeFi is lost to AI-enabled attackers it is unclear where the crypto space goes given DeFi is among the most significant current blockchain applications.