MODELING THE IMPACT OF CYBER THREAT DETECTION SYSTEMS ON THE LEVEL OF CYBER DAMAGE: DID ANALYSIS AND EVENT INVESTIGATION APPROACH

Abstract

The rapid spread of artificial intelligence technologies in the field of cybersecurity is accompanied by a paradoxical effect: AI-based Detection systems increase the efficiency of detecting cyber incidents, but do not guarantee complete prevention of financial losses from cyber attacks. In this regard, a methodological problem arises of correctly interpreting the dynamics of cyber damage after the implementation of intelligent protection systems. The purpose of the study is to econometrically assess the causal impact of the use of artificial intelligence-based cyber threat detection systems on the level of cyber damage. The empirical analysis is based on a data set covering cyber incidents in nine countries around the world (Australia, Brazil, China, France, Germany, India, Japan, the United Kingdom, the United States of America) for the period 2017–2024. The paper proposes an author's cyber damage index, formed on the basis of the integration of financial losses from cyber attacks and the number of affected users. To increase the statistical stability of the distributions, logarithmization of indicators and standardization by Z-score were applied. To assess the causal effect, the Difference-in-Differences econometric approach was used, as well as its dynamic extension in the form of event-study analysis, which allows us to study the time trajectory of the impact of the implementation of AI-based Detection systems. The results obtained indicate that the use of artificial intelligence systems has a statistically significant impact on the dynamics of cyber damage. In the short term after the implementation of AI-based Detection, an increase in the cyber damage index is observed, which is explained by an increase in the level of detection and fixation of cyber incidents that could have previously gone unnoticed. At the same time, in the medium term (3–4 years after implementation), a stable trend towards a decrease in cyber damage is formed, which reflects a gradual increase in the efficiency of AI-systems, their adaptation and integration into the cyber defense infrastructure. The results obtained demonstrate that the implementation of artificial intelligence technologies in the field of cybersecurity changes not only the level of protection, but also the mechanism for measuring cyber risks. The proposed approach allows for a more correct assessment of the effectiveness of AI-based cyber protection systems and is recommended for the analysis of digital security policies and cyber risk management.