Empirical Security Raises $12M to Bring Custom AI to Cyber Defense

Empirical Security has raised $12 million in seed funding to tackle a long-standing challenge in cybersecurity: generic threat models that don’t reflect the unique makeup of individual organizations. Founded by the team behind Kenna Security and the Exploit Prediction Scoring System (EPSS), the company is betting on customized AI as the path forward.

The platform builds dual-layer models, a global model trained on millions of daily exploitation events, and a local model tuned to each customer’s specific infrastructure, systems, and data. The goal isn’t just faster detection, but more accurate prioritization that reflects real-world business risk.

This tailored approach is a shift from traditional risk scoring, where security teams often rely on broad metrics that don’t always map to actual threats. By contrast, Empirical's models are built to support real-time decisions grounded in the context of each environment, helping CISOs back their calls with data that actually fits their operating reality.

This approach also makes Empirical a compelling play for MSSPs. Managed security providers often serve a wide range of clients with vastly different risk profiles. Empirical’s local model architecture allows MSSPs to fine-tune AI for each customer without having to build separate systems from scratch. It’s a scalable way to offer personalized, evidence-backed threat prioritization as a service.

Backed by Costanoa Ventures and a group of seasoned security investors, Empirical is positioning itself as the next evolution in risk-based security, one where local context isn’t an afterthought, but the starting point.