Artificial Intelligence, Violent Extremism and Nihilistic Violence: The Need for Contextual Awareness and Threat-Specific Evaluation in AI
Webinar date: 5 November 2025
14h00-15h30 Johannesburg time; 12h00 London; 07h00 New York
Free for AfATAP and other TAP members, ZAR200 (approx US$13) for non-members
Enquiries: doc@afatap.africa
Presenters: Dr. Kye Allen, Arthur Bradley, Dr. Allysa Czerwinsky, Miron Lakomy, Stuart Macdonald, Broderick James McDonald, and Anda Solea
Abstract:
In this presentation we will examine how violent extremists and other perpetrators of extreme violence are exploiting Artificial Intelligence (AI) tools, drawing on in-depth empirical research across four categories of threat actor: Salafi-jihadism, the extreme right-wing, incels and violent misogyny, and networks associated with nihilistic extreme violence. From operational planning and propaganda production to AI music, voice cloning, ideological roleplay and the translation and summation of materials, we will compare how each category of threat actor utilises AI tools. Our central claim is that there are important differences in terms of these actors’ intent, audience, legitimacy, symbolism, and mobilisation. An appreciation of these distinctions is vital to understand the nuances in how they perceive and use different AI tools.
On this basis, we go on to argue that evaluations of AI safety that are based on broad, capability-centric assessments will miss the mark. Evaluations and safeguards need actor-specific evidence, not just generic misuse categories. Accordingly, we propose four principles: contextual awareness, adversarial robustness, temporal responsiveness, and whole pathway testing. Taken together, these amount to an agenda rather than a checklist: a case for grounding AI safety evaluation in granular, disaggregated and expert-led evidence about specific adversaries, rather than in generalised assumptions about capability. Such an approach has value not just in understanding and responding to extremist and terrorist use of AI, but also to other high-severity adversarial harms, including criminal misuse and hostile state misuse.
Speaker Bios:
Dr. Kye Allen is presently an early career fellow at the University of Cambridge and holds a DPhil in International Relations from the University of Oxford. Alongside his academic post, Kye provides subject matter expertise in political extremism and red teaming services to pattrn.ai (Pattrn Analytics & Intelligence). His current research centres on the prospective links between chatbots and radicalisation processes, alongside the ways in which neo-Nazi communities view AI. More broadly, Kye’s primary research interests focus on the political thought of far-right extremists, both historical and contemporary, and the intersection between technology, political violence, and extremism.
Arthur Bradley is an independent specialist in tracking and analysing terrorist and other malevolent use of online platforms and OSINT Lead at the VOX-Pol Institute. He provides consultancy services to the private sector, non-governmental organisations, IGOs, academia, public sector institutions, and law enforcement agencies. Prior to working independently, Arthur was OSINT manager at Tech Against Terrorism.
Dr. Allysa Czerwinsky is a Research Fellow in AI Trust and Security at the University of Manchester, UK. Her research centres the role manosphere-affiliated communities play in (re)producing male supremacism and misogynist extremism in online environments, as well as how Generative AI can both create and exacerbate online harms, hate-based victimisation, and ideological extremism through synthetic images and videos.
Miron Lakomy is a Professor at the Institute of Political Sciences, the University of Silesia, Poland, and a non-resident fellow at the ITSTIME research centre in Italy. He is an open-source intelligence trainer and investigator with over 15 years of experience in analysing terrorist activities across various layers of online communication. He was involved in projects that included, among others, WEBINT, SOCMINT, IMINT, and GEOINT. During his academic career, he held visiting research positions at the University of Oxford, the University of Cambridge (twice, in 2011 and 2018), the European University Institute, the University of Milan (non-resident), and the Catholic University of the Sacred Heart in Milan (NAWA Bekker program). He has led numerous research grants, including those funded by the Global Network on Extremism & Technology (King’s College London), the International Council for Canadian Studies, and the National Science Centre in Poland. He is a member of the EU Knowledge Hub on Prevention of Radicalisation Research Committee.
Stuart Macdonald is Professor of Law at Swansea University, UK. His research interests lie in criminal law and counterterrorism, particularly extremist and terrorist exploitation of online platforms and regulatory responses. He is Co-Director of the University’s Cyber Threats Research Centre (CYTREC), lead organiser of the biennial #TASMConf (Terrorism and Social Media Conference), Coordinator of the VOX-Pol Network and Director of the VOX-Pol Institute. Stuart has received research funding from the British Academy, Welsh Government, US Government, NATO and the EU, among others. He is a member of the Steering Committee of Europol’s Advisory Network on Terrorism and Propaganda, a Senior Fellow at Hedayah and an Associate Fellow at the International Centre for Counter-Terrorism.
Broderick James McDonald is a Behavioural Scientist at the University of Oxford’s AI Governance Initiative (AIGI) and The Alan Turing Institute (CETaS). Outside of this, he serves as a Senior Advisor to Tech Against Terrorism and a Research Affiliate at the University of Cambridge’s Minderoo Centre for Technology and Democracy. His work examines AI safety and catastrophic risk across online harms and national-security threats, including terrorism, CBRN, hybrid warfare, harmful manipulation, deepfakes, scams and fraud, child safety and election interference. His research and commentary have appeared in The New York Times, The Washington Post, The Wall Street Journal, Foreign Affairs, and The Guardian. He frequently provides expert analysis for international broadcasters, including the BBC, PBS, CBC, ABC News, Good Morning America, France24, and Al Jazeera.
Anda Solea is a Senior Lecturer in Cybercrime and PhD Candidate at the University of Portsmouth, UK, and a Fellow of the Institute for Research on Male Supremacism. Her research explores how extremist ideologies and online subcultures are disseminated and normalised through social media. She specialises in the communication practices of incels, looksmaxxers, and manosphere influencers, with a particular focus on the role of short-form content in the spread of misinformation and propaganda, online radicalisation, and gender-based violence. Her research has been published in the European Journal on Criminal Policy and Research and Crime, Media, Culture, as well as by the Global Network on Extremism and Technology (GNET). Her work has also received significant media attention, including coverage by the BBC and The Guardian.
