Stripe has revealed that its AI-powered fraud detection system, Radar, now blocks approximately $35 billion in fraudulent transactions annually across its platform — a 140% increase from $14.7 billion in 2024.
The improvement is driven by machine learning models trained on Stripe's unique dataset: billions of transactions across millions of businesses worldwide. The scale of data gives Stripe a significant competitive advantage in fraud detection.
False positive rates have simultaneously dropped by 62%, meaning legitimate transactions are less likely to be incorrectly blocked. This dual improvement — catching more fraud while blocking fewer real customers — is the holy grail of payment security.
Stripe processes over $1 trillion in annual payment volume, making its fraud data one of the most comprehensive in the financial system.