Adopting AI, inviting risk? An empirical analysis of corporate cybersecurity consequences in IT firms

Xiao, Wei, Xu, Shan, Li, Jin, Mou, Jian
Electronic Markets · Springer Nature · 2026-10-20 · DOI: 10.1007/s12525-026-00939-3

Аннотация

The rapid advancement of artificial intelligence technologies presents transformative potential alongside unprecedented cybersecurity challenges. Despite widespread industry concerns, empirical evidence establishing this relationship remains limited. Drawing upon Routine Activity Theory, this study examines the impact of artificial intelligence implementation on cybersecurity risks within the information technology sector, and employs an instrumental variable approach to establish causality and address endogeneity. We unpack the framework along three dimensions. First, using a large language model to classify disclosure contexts, we find that only substantive AI implementation, not pre-implementation signaling, raises cybersecurity risk. Second, the risk profile is heterogeneous. We find that frontier AI increases cybersecurity risk, while prior experience with fundamental AI attenuates this effect. Third, for realized breaches, general AI implementation increases externally hacked breaches but not overall breach likelihood, whereas frontier AI raises overall breach likelihood, external hacking, and accidental internal breaches. Malicious insider breaches show no such association. This study provides a causal architecture for understanding artificial intelligence-cybersecurity dynamics, highlighting the necessity for continuous strategic adaptation in the field of cybersecurity and underscoring the importance of striking a balance between artificial intelligence implementation and cybersecurity to ensure a secure and responsible digital ecosystem.

Читать у издателя →

Похожие статьи