Given the diverse backgrounds of people living in modern societies as well as the international nature of cyber-terrorist threats, profiling the type of person behind cyber-mediated crimes has become a norm in terrorist profiling practice. This study contributes to timely efficient terrorist profiling and threat assessment by showcasing an automated content analysis of cyber-mediated terrorist texts, using natural language processing technology and AI-assisted analysis. To characterise the terrorist type of texts and provide clues to threats, the study employs a ‘psycholinguistic profiling’ approach to authorship analysis (Grant 2008). That is, it seeks to describe the likelihood of an author’s engagement in violent extremist activity, identify motives for violence, and provide clues vis-a-vis would-be and actual violent behaviours. The study takes twenty texts produced by international terrorists involved in jihadism and far-right violent extremism as a case study. The findings reveal the investigative value of automated psycholinguistic profiling for security and intelligence practitioners, with the semantic patterns yielding helpful information for an understanding of the criminal nature of terrorist language. Also revealed is the attentional pattern of extremists and their discourse together with clues-based conclusions about text type, as well as ‘warning’ behaviours and motives for aggression which vary according to the authors’ ideological differences.