Successfully Fighting AI-Powered Fraud with Sigma V4.5

Socure argues that rapidly growing AI-enabled fraud, including deepfakes, stolen identities, synthetic “ghost student” accounts, and coordinated identity attacks, has weakened the effectiveness of traditional fraud-detection systems that evaluate risk signals independently. Its newly announced Sigma V4.5 and Sigma Synthetic V4.5 models use transformer-based architecture to assess identity signals simultaneously and in context, combined with expanded fraud-label curation and signals from its cross-industry Identity Graph of devices, IPs, behaviors, and linked identities. Trained on billions of outcomes across Socure’s network and deployed through its RiskOS platform, the company says the models can identify emerging and coordinated fraud patterns more effectively while reducing friction for legitimate customers. In early production results cited by Socure, an adopter processing millions of credit-card applications achieved fraud capture of up to 90% with a 1:1 good-to-bad ratio, while synthetic-fraud detection reached up to 86%; existing users upgrading from V4.0 are projected to see capture gains of up to 20% for identity fraud and 30% for synthetic fraud.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.