Counter-terrorism agencies increasingly leverage artificial intelligence in their operations, with a strong focus on predictive tools that can identify potential individual attackers or attacks before they occur. To be effective, these tools need to analyse open-source and private surveillance data continuously. Despite advancements, there are significant concerns about their effectiveness and the potential legal and societal impacts. While the broader implications of AI technology have been widely discussed in academia, the specific use of preventive person-based AI tools in counter-terrorism remains underexplored. This research examines the AI tool employed by the Israeli Security Agency (ISA) in the Occupied Palestinian Territories from 2015 to 2017 to identify individual attackers before they strike. Through interviews, open-source data, freedom of information requests, and extensive legal proceedings, the study investigates the tool’s creation, design, capabilities, effectiveness, and impact. It proposes methods to enhance the tool's efficiency and reduce potential harm. Utilising critical terrorism studies, the research questions the necessity of such tools even under optimal conditions and suggests alternative applications of AI to mitigate political violence. The research also suggests that at least part of the lure of those AI tools is found in the lessening of moral responsibility they offer counter-terrorism agencies and agents, as difficult moral decisions are allocated to algorithms. This study not only fills a gap in academic literature but is also relevant for AI developers, law enforcement and policymakers.