AI and (De)Radicalisation Interaction Study

Israel/ 6.2 Research Report November 2022

Abstract

This report presents the results of an experimental process focused on some uses of AI technologies within deradicalisation projects in Israeli society. This study was conducted with institutions (government and non-government) centred on assisting people (“backlash populations”) who are suffering from, witnessing and reporting about radicalisation. We examine how state and civic institutions use digital tools (if any) in their work with certain communities and the general public, focusing on the centrality of the connective link between online and offline radicalisation. Attempts to break this link bring different deradicalisation approaches to the surface.
After a short introduction, section 2 offers some socio-political background to put the study into context for three radicalisation set-ups: jihadist terrorism committed by Palestinians and/or Israeli Arabs; Jewish terrorism performed by supporters of far-right-wing racism and religious nationalism; and xenophobia against LGBTQ+ communities, outcast by religious fanatics. The data was based on previous reports, adding the most important contemporary developments. All three have shown increased activity and statements that emphasise the evolving bond between online radical behaviour and offline violent actions. Here we focus on three factors that relate to online/offline radicalisation: recruitment, incitement, and copycat syndrome.
The following section (3) describes the methodology used for this study. It first includes the application of insights from D.Rad reports (D3, D4), followed by semi-structured interviews (“mapping interviews”) with groups working in deradicalisation activities, supplemented with more research from social media, news reports and official data from state institutions. Here we created a six-step process: 1. Data research; 2. Detection of prominent actors; 3. Mapping interviews; 4. Analysing information; 5. AI assessment; and 6. Setting channels for AI solutions.
Section 4 elaborates on the findings revealed in the research, divided into two sub-sections (4.1, 4.2) describing the relevancy of each institution that agreed to share its daily need for, and uses of, technology. Each sub-section presents a summary of data from four institutions that cooperated with and performed the mapping interview process: The Anti-Racism Governmental Unit (GO), FakeReporter (NGO), Ajeec-Nipsed (NGO), and Tag Meir (NGO). Each participant held a key position within the organisation. We found two main channels by which the work on online/offline radicalisation can be established under the I-GAP spectrum framework: preventive and interactive.
Finally, the last section (5) presents the options for adjusting AI solutions in the fight against radicalisation. The preventive route is based on reports that the public submits to the relevant institution, which is slowly becoming a central practice in the field of online/offline radicalisation. Additional digital mechanisms are required since, even if institutions can detect radicalisation, it does not seem yet that there are suitable tools to handle large-scale information. This involves a set of practices and particular means to extract ad-hoc threats and recognise patterns. Developing a local I-GAP lexicon by the relevant institution in Hebrew can be the first step to implementing the detection of users at risk of radicalisation (RoR) (see D6.1). RoR refers to users who spend significant time online and are recognised as highly exposed to incitive content.
The integrative channel shows that civic actors emphasise the connection with RoR and/or backlash populations as part of their daily routine. NGOs’ knowledge can be valuable to the state while developing deradicalisation initiatives. Working with NGOs has shown that many civic initiatives involve educational activities, which can be used as an anchor to future work with young adults from different sectors. Here we find that I-GAP surveys can be used for data collection. AI technologies can produce analysed data reports for working with, and keeping track of, occurrences within backlash populations.