Prediction Market Cheating Gets Creative
An Israeli journalist was threatened for reporting news affecting the payout of a bet on Iranian strikes.
By Alex Goldenberg
Chad CroweMax Meizlish and I argued in these pages in February that prediction markets had become an intelligence asset, vulnerable to insider trading by those with classified knowledge of future events.
Two months later that warning was confirmed in court.
In April, federal prosecutors in New York indicted Master Sgt. Gannon Ken Van Dyke, a special-forces soldier at Fort Bragg, N.C., for using his role in the planning of the January raid that captured Nicolás Maduro to win some $410,000 on Polymarket.
The Justice Department called it the first criminal case in the U.S. arising from prediction-market wagers.
It won’t be the last.
The Van Dyke indictment is the legally easy version of the problem.
Insider misuse of nonpublic information is a well-understood category of misconduct.
Prosecutors charged Mr. Van Dyke with commodities and wire fraud, among other violations, and the platforms cooperated with federal investigators.
Three other recent cases show what the existing framework can’t reach.
In March, a Times of Israel reporter, Emanuel Fabian, posted a routine update to his paper’s live blog reporting that an Iranian missile had struck an open area outside Beit Shemesh, west of Jerusalem.
Within hours, strangers began demanding he revise the story to claim the impact had come from intercepted missile fragments rather than a warhead.
When he refused, the threats began, referencing his family and his home address.
More than $14 million had been wagered on a Polymarket contract titled “Iran strikes Israel on . . .?”, and the contract paid out only on confirmed impact.
Mr. Fabian’s accurate reporting had cost a particular set of bettors their wagers, and they were trying to coerce him into rewriting reality so that the market would resolve their way.
In April, a Polymarket trader walked away with $21,398 on a $119 bet that the temperature in Paris would exceed a given threshold on a particular day.
The bet rested on a single Météo-France weather sensor sitting near the perimeter of Charles de Gaulle Airport on a publicly accessible road.
The sensor recorded a spike of 6 degrees Celsius (nearly 11 degrees Fahrenheit) in 12 minutes before dropping back, and no other station in the area registered the anomaly.
Météo-France filed a criminal complaint, and the leading theory in the press is that someone walked up to the sensor with a battery-powered hair dryer.
A third case shows the threat in its most scalable form.
In March, a Polymarket contract on whether Benjamin Netanyahu would be out of office by year-end climbed roughly 15 points over the course of a week, with roughly $9 million flowing into the market on March 14 and 15.
The trigger was a rumor cycle, amplified by artificial-intelligence-generated content, claiming that Mr. Netanyahu had died.
The rumor was almost certainly not seeded for trading purposes, and it was quickly debunked.
The market moved anyway.
Bettors who were positioned for the spike collected.
The contract didn’t need to resolve their way; it only needed to swing long enough for them to exit.
These three cases describe a new kind of financial manipulation that existing laws aren’t built to handle.
They aren’t insider trading.
They show that prediction-market prices can be moved through attacks on the information environment, the resolution sources, or the people producing the inputs the markets depend on.
The Netanyahu case is the most worrying, because it shows the mechanism working without anyone deliberately targeting the market.
The toxic content, AI-generated rumor cycles and coordinated inauthentic activity that already saturate the information environment are now sufficient, on their own, to move prediction-market prices in profitable directions.
The next iteration won’t be accidental.
The infrastructure for manipulation already exists, refined over the past decade by affiliate marketers, engagement farmers and crypto pump groups—and now intensified by AI.
None of it was built for prediction markets.
All of it can now be pointed at them.
Three steps would help close the gap between the threats to prediction markets and the tools available to combat them.
First, the Commodity Futures Trading Commission should require prediction-market platforms to monitor on-platform activity and the resolution sources their contracts depend on.
Platforms are continuing to build trading surveillance capabilities, but they haven’t done the same for the information that resolves their contracts.
They should be required to publish their resolution sources, document fallback procedures, and monitor markets for upstream tampering and coordinated narrative attacks tied to large open positions.
Second, federal authorities should develop investigative protocols for manipulation campaigns linked to market positions, particularly cases in which harassment or coercion appears tied to large open trades.
The Fabian case is already under investigation by Israeli police.
Journalists and others whose work could resolve a market face similar exposure.
Third, Congress should clarify how existing fraud and market-manipulation statutes apply to conduct designed to move prediction-market prices.
The behavior in the cases above is plausibly illegal under several existing frameworks, but no case has tested those theories against off-platform manipulation of prediction markets.
Clarification would deter the next round of manipulation attempts and give regulators a workable tool before the problem scales.
Prediction markets aren’t going away, and their trading volume is growing.
The first wave of cases involved insiders betting on classified information they weren’t supposed to share.
The next wave will involve outsiders manufacturing the information itself.
Mr. Goldenberg is the founder of Silent Index, a national security consultancy, and a senior fellow at Rutgers University’s Miller Center.
0 comments:
Publicar un comentario