5 steps for frontier AI readiness

The report warns that frontier AI models are accelerating vulnerability discovery and exploit development, compressing the time between exposure and exploitation and undermining traditional vulnerability management. It recommends shifting to exposure-focused risk measurement and operationalizing five controls—measuring exploitability, continuous internal/external exposure validation, identity-centric defenses, machine-speed detection and response, and governed AI usage—to reduce attack surface and improve resilience.