EVALUATION OF AI ATTACK MITIGATION FROM CITRIX BLEED TO SELF-EVOLVING MALWARE: MODERNISING AEROSPACE CYBER DEFENCE WITH AI
Daniel Schönle, Christoph Reich · 2025
This study evaluates the effectiveness of cybersecurity architectures in mitigating AI-driven cyberattacks. A key tool in this evaluation is the Layered Rubric Security Score (LRS), a structured threat model-based MITRE ATT&CK framework that has been developed specifically for this purpose. LRS utilises architectural layers, cross-detection, containment, and automated response capabilities for assessment. Modern reference architectures from aerospace enterprises, designed to mitigate attacks like Citrix Bleed and RedLine Stealer, are assessed and compared. Findings reveal significant gaps in handling adaptive and LLM-based threats, particularly in the integration of CASB and EDR/XDR. LRS introduces a scoring methodology and proposes mitigation strategies aligned with Zero Trust Architecture, supporting the evidence-based improvement of AI-resilient defence infrastructures.