Socure has launched Socure Signals, giving enterprises direct access to the feature store powering Socure’s market-leading ML models and the industry’s most accurate identity and risk decisions. By opening access to the same data attributes, probabilistic features, and graph consortium intelligence that fuel Socure’s fraud and identity models, enterprises gain the transparency and flexibility to fine-tune strategies, tailor machine-learning models to their own problem set, and outpace complex AI-powered fraud schemes. Socure Signals introduces three categories of intelligence, each capturing a different layer of risk: Input-Derived Signals: Data attributes from PII, documents, biometrics, and behavioral interactions, including phone carrier type, email domain age, ID barcode fields, selfie liveness, and click-path behavior. Model-Derived Signals: Advanced, probabilistic features from Socure’s proprietary AI models, such as perplexity scores for randomized identifiers, as well as computer vision variables that detect subtle document tampering, injection attempts, or presentation attacks. Graph-Derived Signals: Insights from Socure’s proprietary Identity Graph, linking billions of entities, devices, and behaviors across industries to expose cross-institution fraud rings, velocity anomalies, and consortium-level first-party fraud patterns. Delivered via API alongside Socure’s model risk scores and reason codes, these signals give organizations the precision and transparency to understand both why a score was returned and the attributes driving it. Within Socure’s RiskOS™ platform, organizations can integrate Signals into custom rules, adaptive workflows, and decisioning paths — enabling fraud teams to respond to emerging threats in real time. In pre-launch testing, a top-tier fintech used phone-specific signals to uncover systemic fraud concentrated within a handful of phone carriers, each showing fraud rates above 90%. By creating custom rules to step up verification for applicants tied to these carriers, the organization was able to disrupt coordinated fraud ring activity and prevent additional losses before they scaled.