Fortisec
Launch ORI
A living research roadmap for operational incompleteness, observability gaps, adversarial telemetry shaping, and uncertainty-aware cyber risk inference.
It establishes that operational incompleteness exists and matters in AI-native cybersecurity. It is intentionally conceptual: it opens the research direction without overloading the first paper with implementation, architecture, metrics, or mathematical machinery.
The foundational paper establishes the central question behind Fortisec Research: whether AI-native defensive systems can reason reliably when the operational environment itself is only partially observable.
The research program builds from that thesis toward more operational work on observability gaps, adversarial telemetry shaping, and uncertainty-aware cyber risk inference.
Establishes the core thesis: observable operational behavior may not fully characterize hidden operational structure under adversarial uncertainty.
Focuses on telemetry incompleteness, cloud visibility fragmentation, AI reasoning boundaries, identity graph gaps, and operational uncertainty propagation.
Investigates how adversaries may shape telemetry so AI systems become more confident while becoming more wrong.
Develops the machinery for Bayesian scoring, observability confidence, hidden-state estimation, risk residual modeling, and probabilistic operational inference.
Reasoning about what defensive systems can and cannot see across telemetry, identities, cloud services, integrations, and dependencies.
Understanding how adversaries exploit ambiguity, incomplete state, synthetic normality, and confidence errors in AI-assisted security workflows.
Developing uncertainty-aware models for hidden exposure, operational confidence, and structurally unobserved cyber risk.