Ethical Capital and Cybersecurity Behaviour: Exploring Large Language Model (LLM) Use in Organizations

Abstract

This study examines how ethical orientations and organizational factors shape users’ ethical cybersecurity behaviour when interacting with large language models (LLMs). Using a quantitative design supported by thematic insights, it investigates how ethical capital and responsible LLM use influence cybersecurity ethics. Findings show that ethical capital affects behaviour primarily through responsible use of LLMs rather than through direct effects. Demographic factors, including age, gender, discipline, education, country, and usage frequency, did not significantly predict behaviour, suggesting that the rapid evolution of LLMs places users from diverse backgrounds on a similar learning curve. Complementary qualitative analysis highlights that human behaviour, organizational culture, and governance structures are as critical as the technology itself. Themes such as output oversight, responsible prompting, data protection, and overreliance reveal persistent human-centred vulnerabilities and the need for adaptive governance. Overall, the study identifies responsible LLM use as a key behavioural pathway for strengthening cybersecurity ethics and underscores the importance of continuous oversight, inclusive training, and human accountability in LLM deployment.