Structural monitoring of attack surface growth through adversarial adaptation, tracking jailbreak-driven expansion.
Attack surfaces in complex systems are not static — they expand through adversarial adaptation. As defenders patch vulnerabilities and harden systems, attackers adapt their techniques to find new attack paths, and the interaction between offense and defense structurally reshapes the attack surface over time. The structural problem is that this co-evolutionary dynamic tends to expand the attack surface faster than defense can contract it, because each defensive change creates new structural interfaces that may introduce new vulnerabilities.
In AI systems specifically, jailbreak research and adversarial prompt engineering create a rapidly evolving attack surface where new techniques continuously expand the space of effective attacks. The structural coupling between defense adaptations and attacker innovation creates an expanding frontier.
This application addresses the dynamic security posture of complex systems under adversarial pressure. The relevant system boundary includes the system's attack surface, the adversarial community's adaptation patterns, the defensive modifications made in response, and the structural co-evolution between offense and defense.
Static security assessment provides a snapshot that becomes obsolete as adversaries adapt. Structural tracking of attack surface evolution provides the dynamic view needed for security strategies that account for adversarial adaptation — the only approach that maintains security posture over time rather than merely at the point of assessment.
The SORT framework addresses this application through four structural dimensions, each providing a distinct analytical layer.
Attack surface grows through adversarial adaptation.
Jailbreak adaptations couple to surface expansion.
Structural monitoring of attack surface evolution.
Surface minimization, adaptation tracking, security posture.