Empirical Security competes by letting enterprises measure and defend the exploitation probabilities its models produce rather than relying on a generic severity ranking. Its platform supports a foundation model that captures attacker behavior across the internet and radiant models trained on an environment’s telemetry; the two work together to surface vulnerabilities likely to be exploited in an infrastructure.
For enterprises, the local model trains on their asset, control, and patching data, while the platform trains its models on exploitation telemetry. CVSS severity scores and EPSS global exploitation probabilities do not identify what will be exploited in a specific environment; Empirical Security publishes a false positive rate for its models and integrates with existing asset inventories, including CMDB, EDR, cloud APIs, scanners, and container registries.
