Tag: XAI

xai-blog

XAI as a Double-Edged Sword: When Explainability Becomes an Attack Surface

2026-10-05
Future AI-native 6G engineering should not treat explainability and cybersecurity as separate disciplines. They are increasingly interconnected. Various XAI methods can illuminate a model's decision-making process. Recent findings on AI-based wireless systems suggest that explainability should not be considered only a tool for understanding AI - it should also be included in the security threat model. In this blog post, we discuss how explanations can improve transparency and trust in AI systems, while also examining the security risks they may introduce. We present an AI–XAI–Security architecture, bringing together AI, XAI, and Security layers, and show how their interaction can reveal both new opportunities and new vulnerabilities in intelligent wireless systems.
Continue reading