Julia Kupper | University of Westminster (original) (raw)

Publications by Julia Kupper

Accelerationism Research Consortium, 2024

This article provides an overview of the propaganda authored and disseminated by the Terrorgram c... more This article provides an overview of the propaganda authored and disseminated by the Terrorgram community, and includes discussions on how Terrorgram publications encourage lone-actors to mobilize to violence; how the products align with and differ from targeted violence manifestos and live-streams; and what the operational implications are for threat assessors, investigators and intel analysts.

Global Network on Extremism and Technology, 2023

This pilot study investigates the correlation of online and on-the-ground behaviours of three lon... more This pilot study investigates the correlation of online and on-the-ground behaviours of three lone-actor terrorists prior to their intended and planned attacks on soft targets in North America and Europe: the Pittsburgh synagogue shooter, the Buffalo supermarket shooter and the Bratislava bar shooter. The activities were examined with the definition of the proximal warning indicator energy burst from the Terrorist Radicalization Assessment Protocol (TRAP-18), originally defined as an acceleration in frequency or variety of preparatory behaviours related to the target. An extensive quantitative and qualitative assessment of primary and secondary sources was conducted. This included raw data from different technology platforms (Gab, Discord and Twitter – now X) and open-source materials, such as criminal complaints, superseding indictments and court trial transcripts.

Preliminary findings from this small sample suggest an inverse relationship between online and offline behaviours across all three perpetrators. The average length of time between the decision to attack and the actual attack was five months, with an elevation of digital activities in the three months leading up to the incident, along with some indications of offline planning. In the week prior to the event, social media activity decreased – with two subjects going completely dark on the day before the respective acts of violence – while terrestrial preparations increased. On the actual day of the incident, all assailants accelerated their tactical on-the-ground actions and resurfaced in the online sphere to publish their final messages in the minutes or hours prior to the attack. It appears that the energy burst behaviours, in the digital sphere and in offline actions, can be measured in both frequency and variety. Operational implications of this negative correlation are suggested for intelligence analysts, counterterrorism investigators and threat assessors, and policy recommendations are proposed for technology companies and policymakers.

Accelerationism Research Consortium, 2023

On October 12, 2022, 19-year-old Juraj Krajčík perpetrated a terrorist attack that killed two peo... more On October 12, 2022, 19-year-old Juraj Krajčík perpetrated a terrorist attack that killed two people and injured one outside a LGBTQ+ bar in the center of Bratislava, Slovakia. Krajčík was deeply motivated by the militant accelerationism movement, Terrorgram (a prominent space within Telegram's far-right ecosystem), and the so-called Saints Culture. Krajčík’s attack and manifesto now exist as the clearest exemplar of an accelerationist attacker since Brenton Tarrant.

Journal of Threat Assessment and Management, 2023

This case study is a detailed assessment of the paranoid schizophrenia and right-wing extremist b... more This case study is a detailed assessment of the paranoid schizophrenia and right-wing extremist beliefs of Tobias Rathjen, who amalgamated three low-base rate events: mass homicide, matricide, and suicide. The offender killed nine individuals during a terrorist attack in Hanau, Germany, on February 19, 2020, before murdering his mother and taking his own life. A comprehensive qualitative analysis of primary and secondary sources was conducted, with raw materials consisting of three written documents and three recorded videos, all authored and disseminated by the perpetrator prior to his attack. Open-source data were predominantly composed of the postmortem forensic psychiatric evaluation of the assailant, as well as minor interpellations between the German government and parliamentary groups, and to a lesser extent printed and online articles. We structured the analysis with the Terrorist Radicalization Assessment Protocol (TRAP-18) and found that Rathjen was positive for 94% of the TRAP-18 indicators. The motivations for his acts of violence were multifaceted, but his major psychiatric illness, paranoid schizophrenia with chronic delusions, was central in his progression to become a lone-actor terrorist. In addition to this internally formed psychotic-driven ideology, the perpetrator was influenced and inspired by external (i.e., online and offline) xenophobic and conspiratorial elements that gradually provided a secondary framework for his violent attack. The complex interplay of his delusions, obsessions, and extreme overvalued beliefs-which drove his fixation-reveal the difficulty of clinically understanding such a case, and also the necessity of doing so, to attempt to risk mitigate these types of subjects by threat assessment teams.

Global Network on Extremism & Technology

This Insight provides an overview of the ideology behind the German Reichsbürger movement, its re... more This Insight provides an overview of the ideology behind the German Reichsbürger movement, its recent cross-pollination with QAnon narratives, and how this amalgamation resulted in the planned German coup by a group called the ‘Patriotische Union’ (the ‘Patriotic Union’). Furthermore, we emphasize the conspiratorial belief systems of several arrested members and discuss why secure communications platforms, such as Telegram, are perpetually exploited to spread extremist content and coordinate acts of violence.

Bulletin

This article argues that written and spoken communications authored and leaked by lone perpetrato... more This article argues that written and spoken communications authored and leaked by lone perpetrators like Cano who actively plan and prepare targeted violence can be utilized to prevent destructive attacks. Targeted violence refers to intended attacks by an offender who preselects one or more targets, such as people at a specific location in a public setting (e.g., school, workplace, concert).

Perspectives on Terrorism, 2022

This article corroborates the continued threat of extreme right terrorism by exemplifying textual... more This article corroborates the continued threat of extreme right terrorism by exemplifying textually interconnected links across linguistic evidence composed prior to or during attacks in the United States, New Zealand, Germany, Norway and Sweden. A qualitative content analysis of targeted violence manifestos and live-streams, attack announcements on online platforms, and writings on equipment (e.g., firearms) used during the incidents reveals an emerging illicit genre set that is increasingly consolidated in form and function. The messages accentuate an intricate far-right online ecosystem that empowers copycats and escorts them on their pathway to violence. A definition for targeted violence live-streams is proposed and operational applications are discussed.

IEEE International Conference on Big Data, 2021

Suicide is a global health problem with more than 700,000 individuals dying by self-destruction e... more Suicide is a global health problem with more than 700,000 individuals dying by self-destruction each year, yet it is classified as a low base rate behavior that is difficult to prognosticate. Aiming to advance suicide prediction and prevention, we examined the potential use of machine learning and text analyses models to predict suicide risk based on written communications. Specifically, we used a dataset consisting of more than 27,000 general writings unrelated to suicide, 193 genuine suicide notes from individuals who committed suicide, and an additional 89 suicide posts shared on sub-Reddits for an in-the-wild test to examine the prediction accuracy of two machine learning models (SVM & RoBERTa) and a linguistic marker model. Our tests showed that the machine learning models performed better than the linguistic marker model when examined on the test data. However, the linguistic marker model achieved higher results in the wild, correctly classifying 88% of written communications as a "high risk of suicide" versus 56% and 70% of the machine learning models. The best in-the-wild performing model was adopted in an online suicide risk assessment tool called Edwin to honor Edwin Shneidman for his numerous contributions to the field of suicidology. Finally, discrepancies between training and real-world data, vocabulary variation across domains, and the limited number of benchmarks constitute limitations that need to be addressed in future research.

Journal of Threat Assessment and Management, 2021

Julia Kupper & Dr. Reid Meloy - https://doi.org/10.1037/tam0000165 This study is a qualitative... more Julia Kupper & Dr. Reid Meloy - https://doi.org/10.1037/tam0000165

This study is a qualitative and quantitative analysis of written and spoken manifestos authored by lone offenders that planned to or committed a targeted attack (n = 30). The acts of violence were primarily motivated by an ideology or a personal grievance, and occurred between 1974–2021. The main objective of this retrospective study was to examine if a behavior-based threat assessment instrument, the Terrorist Radicalization Assessment Protocol (TRAP-18), could be applied to a thin slice of data using only language evidence compiled by the perpetrator before the event occurred. Findings indicate that 17 out of 18 TRAP-18 indicators (94%) were able to be coded in the manifestos. Proximal warning behaviors ranged from 2 to 7, with an average of 4.5. The average number of distal characteristics was 3.8 across the sample. The most frequent proximal warning behaviors were leakage, identification, fixation, and last resort. The most frequent distal characteristics were changes in thinking and emotion, framed by an ideology, and personal grievance and moral outrage. Further analyses of the manifestos comparing written versus spoken communications, ideologically based versus grievance-fueled attacks, and seven categories of primary motivations, supported the generalizability of the TRAP-18 when applied to only language data. A definition for targeted violence manifestos is proposed and examined within a genre framework. Application to threat assessment and management is discussed.