Security testing has traditionally worked in snapshots: test, find the problems, fix them, and test again later.
But this model no longer works in an AI-driven environment changing at unprecedented speed. Organisations need to evolve their defenses just as fast.
AI is accelerating changes to the attack surface while giving attackers greater speed and capability. A point-in-time test can therefore become outdated almost as soon as it is completed.
In a recent open letter, AI leaders including OpenAI, Anthropic, Google and Microsoft warned that AI-enabled attacks will become far more widespread and sophisticated, and called for defenses to be continuously tested against the most advanced AI-enabled cyber threats. Sandro Nafzger explores why traditional testing is becoming outdated, and why the future lies in intelligent security testing: testing what matters, when it matters, and learning from every cycle.