AI Terrorism Is Just Terrorism

The Atlantic | Ian Bogost

Artificial intelligence has prompted renewed warnings about terrorism, but Ian Bogost questions whether the technology fundamentally changes the threat. Governments, international organisations and research institutions have examined possible terrorist uses of AI, including cyberattacks, autonomous vehicles, facial-recognition spoofing, deepfakes and engineered biological weapons. The debate has also extended to language models and their potential influence on propaganda and radicalisation.

The United Nations Office of Counter-Terrorism co-published a report on the malicious use of artificial intelligence for terrorist purposes in 2021. The Middlebury Institute of International Studies published two reports in 2019 and 2020 examining risks associated with language models. Bogost also describes a Harvard Kennedy School working paper that treats a major AI-enabled terrorist event as a credible possibility. He places these concerns within the unsettled history of terrorism as a concept, from state violence during revolutionary France to anarchist violence and later separatist and ideologically motivated attacks. Cyberterrorism further complicates the definition because digital disruption can involve state actors and may occur without physical violence. The available extract questions whether AI represents a fundamentally new category of threat.

Comment

The central analytical problem is classification: whether AI creates a distinct form of terrorism or changes the means through which existing terrorist objectives are pursued. A technology-centred label can obscure differences between propaganda, cyber intrusion, physical attacks and the use of automated systems. These activities have different mechanisms and consequences, even when they share an intended psychological effect. Bogost's discussion of cyberterrorism illustrates how the label itself can blur the distinction between politically motivated violence and digital disruption.

The history outlined in the extract shows why definitions matter. During revolutionary France, terrorism referred to violence exercised by the state; by the late nineteenth century, the term had become associated with non-state anarchist violence. That shift demonstrates that the concept changes with the actors and political context under examination. Applying the same label to AI-assisted propaganda, infrastructure intrusion and physical attacks risks treating unlike threats as equivalent before their effects are established.

The practical analytical distinction is between a new capability and a new category of violence. The United Nations Office of Counter-Terrorism report cited by Bogost identifies malicious AI use as a security concern, but the existence of possible applications alone does not establish a fundamentally new terrorist phenomenon. The supplied extract leaves open whether AI changes the underlying character of terrorism or mainly expands the tools available to existing actors.

Strategic Question for Discussion

What evidence would establish that AI-enabled terrorism represents a distinct threat category rather than a technological extension of established terrorist methods?

The distinction depends on demonstrated changes in capability, scale or the relationship between an attack and its intended effect. The United Nations Office of Counter-Terrorism report establishes that malicious AI use has received institutional attention, but concern alone does not prove a new category of terrorism. The available extract leaves this question unresolved.

Share your assessment in the comments below.

To read in details click here