Update on Fortinet Use of Frontier AI

Fortinet describes integrating frontier AI models (e.g., Mythos, GPT 5.5 Cyber) into their security testing pipeline to analyze firmware and source code at scale; results showed high-quality AI-generated threat models, effective firmware reverse-engineering and proof-of-concept generation, and source-code findings with ~40% true positives, ~30% false positives, and ~30% security best-practice suggestions. The report warns that frontier models accelerate reconnaissance and weaponization (reducing time-to-exploit), recommends shortening mitigation cycles (virtual patching, secure-by-default deployments, MFA, CTEM, automated orchestration), and commits to addressing discovered issues within their SDLC.