How a Russian Gang Used American Companies To Create a Fraud Empire
324 defendants, $14.6 billion in attempted fraud, and one Russia-linked ring responsible for more of it than all 300-plus other defendants combined. The real money in insurance fraud has never been in the opportunist filing one bad claim. In this article we describe the Fraud, and explore how AI can be used to fight it.
Insurance fraud is a silent tax on everyone. Workers comp claims with erroneous fake injuries, crash for cash schemes involving intentional automotive accidents, these drive up the cost of insurance in every industry, the expense of which trickles down to the customer. Insurance Fraud is estimated at $308.6 billion annually in the US, with global estimates of insurance fraud surpassing $1 Trillion.
The vast majority of fraudulent insurance claims, by volume, are filed by individuals. Opportunistic fraud (individuals inflating or exaggerating a legitimate claim) accounts for 44% of tracked incidents. Combined with individual fake injuries (29%) and rate evasion (9%), individual-driven activity encompasses over 80% of all flagged cases, while organized crime rings account for roughly 13% of tracked cases by volume (source). If one tracks fraud by the number of claims, one is left to conclude that most fraud is performed by opportunistic individuals. If one tracks fraud by the amount of money defrauded, however, a new story emerges.
In the Justice Department’s 2025 National Health Care Fraud Takedown, 324 defendants were charged in schemes that attempted to steal a combined $14.6 billion. One group accounted for $10.6 billion of that total. They used the stolen identities of more than a million Americans to bill Medicare for catheters and other equipment. They procured these stolen identities by buying legitimate businesses, then used the billing information stored in these businesses to create profiles that they used to impersonate Americans.
These statements are allegations, based on Department of Justice, and have not been proven in a court of law.
In this case, one organized network was charged with more fraud than the other 300+ defendants combined. The optics of individuals awkwardly staging slip-and-falls on security cameras make for good television, but it’s the coordinated groups of professionals actively engaged in coordinated schemes that move real money.
There’s been a cat-and-mouse game between organized crime and fraud investigators for hundreds of years; organized rings were intentionally sinking ships to defraud insurance companies over 2,000 years ago. Across this entire time, detecting fraud has been a labor-intensive effort, a fact which fraudsters have consistently exploited. One peer-reviewed analysis found roughly 99.9% of claims are resolved and treated as legitimate without ever being examined (source), meaning fraudsters don’t have to be very sophisticated to get away with fraud, they just need to stick their needle in the haystack.
With AI, this dynamic is starting to flip. In 2025, the general sentiment was that AI powered systems would create significant job displacement. In 2026, many businesses are realizing that AI’s true power isn’t that of replacing humans, but making them more productive. Fraud investigation is exactly the type of work where support is most needed, and it’s resulting in significant speedups.
FraudX, an AI-powered fraud detection platform, has allowed fraud investigators to speed up operations significantly.
“With FraudX, we're combining AI with legal and investigative strength to protect our clients, accelerate legitimate claims and safeguard the integrity of insurance systems across the marketplace”
For many, this has been a turnkey improvement that’s increased coverage by several orders of magnitude.
“Our customers face millions of pages of claims. With FraudX, we help them turn months of work into minutes”
FraudX speeds up fraud investigation teams in a variety of ways. One is by accelerating lead qualification, which addresses the problem that most fraudulent claims are never even considered by investigators. The platform can read through thousands of pages of documents in minutes, highlighting discrepancies and red flags. FraudX then uses those red flags to calculate a “fraud score” for each claim. After FraudX analyzes a set of claims, investigators can see a ranked list of those claims ordered by likelihood of fraud, allowing them to prioritize the claims most worth their attention. FraudX also handles other labor-intensive parts of the job: searching large document sets for specific information, scanning for the names of known bad actors, and identifying trends that persist across many claims.
As systems like FraudX continue to develop, fraud investigators will have the tools they need to examine many more claims, much more closely. If you’re interested in detecting fraud faster and more accurately, schedule a demo, and we’ll show you how FraudX can accelerate how you find fraud.