Lone-actor terrorism is a major threat to national security in both Europe and North America. TRAP-18 is a risk assessment protocol based on a psychological framework, specifically developed for lone-actor terrorists. Digital forums have shown to be an increased platform for radicalization and communication of future attacks. Akrami et al. (2022) have developed a digital risk assessment measure used for identifying potential future lone actors in identifying Person of Security Concern (PoSC). The measure uses dictionary-based text analysis and machine-learning techniques that classify cases by a General Risk Score (GRS). The present study is a systematic review of the cases that are classified as false-positive by the GRS. A content analysis was performed with a human evaluation to locate the false-positive cases. A thematic analysis was performed and found three themes as the main reason for the false-positive classifications; (a) Mentioning violence or death from a third-party perspective, (b) Depression and suicidal thoughts, and (c) Describing fictional/historical events. Thus, the numan evaluation reduced the false-positive cases by more than 80 percent. THese findings are discussed in relation to the strenghts and weaknesses of digital and manual assessment of risk for violence.