The rapid emergence of generative artificial intelligence has sparked a wave of warnings about its potential misuse by terrorists. From AI-generated propaganda and deepfakes to fears of automated radicalization and AI-designed weapons, policymakers, security agencies, and researchers have increasingly debated whether the technology could trigger a dramatic shift in the nature of terrorism. Yet a new study suggests that many of these concerns may be running ahead of the evidence.
Published in CTC Sentinel, a flagship publication of the Combating Terrorism Center, the analysis examines whether generative AI truly represents a revolutionary advance for terrorist organizations or whether contemporary security debates have exaggerated its practical impact. The study concludes that while AI tools can improve the efficiency and accessibility of certain terrorist activities, there is little indication that they fundamentally change the nature of terrorism or dramatically increase operational capabilities.
The question arrives at a moment when artificial intelligence is rapidly reshaping everyday life. Large language models, image generators, and other AI systems have become widely available to millions of users worldwide. As governments and businesses race to adopt the technology, concerns about malicious use have inevitably followed. Terrorism analysts, long accustomed to studying how extremists exploit new technologies, have increasingly turned their attention to AI’s darker possibilities.
History offers good reasons for such scrutiny. The evolution of modern terrorism has often been intertwined with technological change. The invention of dynamite in the nineteenth century provided militants with unprecedented destructive power. The spread of telegraph networks and mass printing allowed radical groups to communicate more effectively and disseminate their messages. In the twentieth century, commercial aviation created new high-profile targets while television amplified the psychological impact of attacks. The internet, social media, and smartphones later enabled extremist organizations to recruit, coordinate, and propagate their ideologies on a global scale.
Against that backdrop, many analysts have wondered whether generative AI represents the next major leap in terrorist capability.
To answer that question, the study draws on the concept of “affordance theory,” a framework that examines how people use technologies in ways that may differ from their original purpose. The idea is simple: technologies possess certain possibilities that users can discover and exploit, often in unexpected ways.
One frequently cited example comes from the conflict in Northern Ireland. A Swiss-made parking timer designed to remind motorists when their parking sessions expired was adapted by the Provisional Irish Republican Army as a reliable and inexpensive triggering mechanism for improvised explosive devices. The technology was not created for violence, but militants identified an unintended use that served their objectives.
According to the study, terrorists have historically excelled at discovering such hidden affordances. Rather than inventing entirely new technologies, they often repurpose existing tools that are cheap, accessible, and reliable. The central question, therefore, is not whether generative AI can theoretically be used for terrorism but whether it provides capabilities that significantly alter the threat landscape.
Much of the current debate focuses on four areas where AI could potentially benefit extremist actors: propaganda, recruitment and radicalization, reconnaissance and research, and attack planning.
In the field of propaganda, generative AI appears to offer the clearest advantages. Terrorist organizations can use AI systems to produce text, images, videos, translations, and other forms of content quickly and at minimal cost. Deepfake technologies raise concerns about realistic but fabricated videos, while automated content generation could allow extremist groups to flood online spaces with messaging at unprecedented scale.
Researchers acknowledge that these capabilities are real. Studies have shown that AI models can generate extremist content with relatively little effort, even when developers implement safeguards designed to prevent misuse. Techniques known as prompt engineering and jailbreaks can sometimes bypass restrictions and produce outputs that developers never intended.
Yet the study argues that the significance of these capabilities may be overstated. While AI can accelerate content production, quantity does not automatically translate into influence. The ability to create more propaganda does not guarantee that more people will be persuaded by it.
This distinction becomes particularly important when examining fears surrounding recruitment and radicalization. Some analysts have warned that AI chatbots could function as virtual recruiters, engaging vulnerable individuals around the clock and gradually steering them toward extremist beliefs. Others have suggested that aspiring terrorists could create personalized digital mentors to reinforce radical ideologies.
One frequently cited example involves Jaswant Singh Chail, who was arrested at Windsor Castle in 2021 carrying a crossbow and later revealed that he had discussed his plans with an AI chatbot he regarded as a companion. The case has often been cited as evidence of AI’s potential role in radicalization.
However, the authors argue that such examples remain exceptional and do not demonstrate a broader transformation in terrorist recruitment. Radicalization, they note, remains a deeply human process shaped by social relationships, personal experiences, ideology, identity, and context. Technologies may facilitate certain aspects of that process, but there is little evidence that AI can automate or replace the complex pathways that lead individuals toward violence.
The study similarly questions claims that AI will revolutionize terrorist reconnaissance and attack planning. Large language models can undoubtedly assist users in gathering information more efficiently than traditional search engines. They can summarize large quantities of data, identify patterns, and provide detailed responses to complex queries.
For terrorist organizations conducting research on potential targets, these tools could reduce the time required to collect and process information. However, researchers note that such capabilities largely represent efficiency gains rather than entirely new possibilities. Terrorists have long conducted remote reconnaissance using publicly available information, satellite imagery, online maps, and social media content. AI may streamline those activities, but it does not fundamentally alter them.
Moreover, AI systems remain vulnerable to inaccuracies and hallucinations—instances where models confidently generate incorrect information. Such errors could create operational risks for users who rely too heavily on AI-generated analysis.
Perhaps the most alarming scenarios involve the possibility that AI could help terrorists develop novel weapons, biological agents, toxins, or sophisticated cyberattacks. Popular discussions often focus on the fear that a determined extremist might use AI to design a devastating new attack method previously beyond their reach.
The study urges caution regarding such predictions. While AI can assist with information gathering and technical explanations, there is currently little evidence that it enables users to leap from curiosity to operational capability in highly specialized fields. Developing effective chemical, biological, or radiological weapons requires extensive expertise, resources, testing, and infrastructure. These barriers have historically prevented most terrorist groups from acquiring such capabilities despite decades of effort.
The researchers point to the example of al-Qaida, which invested significant resources in unconventional weapons research during the 1990s and early 2000s but never succeeded in developing a major operational capability in this area. The challenges associated with such weapons remain substantial, regardless of advances in AI.
One of the study’s most important arguments concerns the difference between capability and effect. Even if AI enables extremists to produce more content, conduct faster research, or explore new ideas, the ultimate question is whether these activities achieve meaningful results.
Communication theory suggests that influence is rarely straightforward. Human beings do not passively absorb information or automatically adopt new beliefs simply because they encounter persuasive content. People filter messages through existing beliefs, social networks, and personal experiences. They often resist persuasion and selectively engage with information that confirms their preexisting views.
This insight challenges assumptions that AI-generated propaganda will inevitably trigger mass radicalization. While extremists may be able to generate enormous quantities of content, they face the same problem confronting governments, corporations, activists, and media organizations: attracting and retaining attention in an increasingly crowded information environment.
The study notes that the rise of AI-generated content may actually intensify competition for attention rather than strengthen extremist influence. As machine-generated material floods digital platforms, audiences may become more skeptical, less engaged, and more difficult to persuade. The growing prevalence of what critics call “AI slop”—low-quality automated content—could dilute the impact of extremist messaging rather than amplify it.
Beyond the technological debate itself, the authors warn against the dangers of exaggerating emerging threats. Security institutions have a long history of focusing on worst-case scenarios involving new technologies, sometimes with unintended consequences.
The fear that terrorists might acquire weapons of mass destruction shaped major policy decisions in the early twenty-first century, including arguments used to justify the 2003 invasion of Iraq. Similarly, predictions of catastrophic cyberterrorism have circulated for decades despite the absence of a major destructive cyberattack conclusively attributed to a terrorist organization.
According to the study, concerns about AI-enabled terrorism may represent the latest chapter in this pattern. While examining potential threats is necessary, excessive focus on speculative dangers can divert attention and resources from more immediate and probable risks.
The researchers argue that governments must carefully balance preparedness with evidence-based assessment. Every dollar spent preparing for hypothetical AI-driven terrorist scenarios is a dollar unavailable for addressing other security challenges. In some cases, threats posed by state-sponsored cyber operations or organized criminal networks may be more likely and more consequential than those involving terrorist misuse of AI.
The analysis ultimately concludes that generative AI should be viewed less as a revolutionary force and more as another tool that terrorists may adapt to existing purposes. Like previous communication technologies, AI offers opportunities for propaganda, outreach, and information gathering. It may lower barriers, increase efficiency, and reduce costs. But there is little evidence that it fundamentally changes what terrorism is or how it operates.
History suggests that terrorists rarely gravitate toward the most sophisticated technologies available. Instead, they consistently favor methods that are simple, accessible, inexpensive, and effective. The deadliest attacks of recent decades have often relied on ordinary tools—vehicles, knives, homemade explosives, and commercially available technologies—rather than cutting-edge innovations.
The study does not dismiss the possibility that AI will play a larger role in future terrorist activities. Rather, it argues that researchers and policymakers should resist the temptation to mistake hypothetical possibilities for demonstrated realities. Imagination remains essential in anticipating emerging threats, but effective counterterrorism requires grounding those concerns in evidence rather than speculation.
As governments around the world grapple with the implications of artificial intelligence, the debate over terrorism highlights a broader challenge. The most important question may not be what AI could theoretically allow extremists to do, but what they are actually doing with it today. So far, the evidence suggests that generative AI is making some terrorist activities easier—but not necessarily making terrorism itself fundamentally different.
