Among these, the company debuted an AI foundation model to improve fraud detection and authorization rates. Dubbed the Payments Foundation Model, it is trained on tens of billions of transactions and incorporates hundreds of “subtle signals” per payment, which it said specialized models cannot capture. The technology firm plans to deploy this model across its payments suite to improve performance in ways that were previously unattainable. Stripe stated that early results indicate the model’s effectiveness, particularly against card testing attacks, where it increased detection rates beyond the 80% reduction achieved over two years with previous models. By applying the new foundation model, Stripe increased its detection rate for attacks on large businesses by 64% practically overnight. In parallel, Stripe expanded its money management offerings with the launch of Stablecoin Financial Accounts, powered by stablecoins. Businesses using these accounts can hold balances in stablecoins, receive payments via both crypto and traditional fiat rails such as ACH and SEPA, and send stablecoins to most markets globally. These accounts are designed to be accessible to businesses in 101 countries. Initially, the accounts will support stablecoins USDC and Bridge’s USDB, with plans to incorporate additional currencies over time. Stripe also announced a deeper partnership with NVIDIA, which completed the fastest-ever migration to Stripe Billing.