'Lavender': The AI machine directing Israel's bombing spree in Gaza (www.972mag.com)

The Israeli military has deployed an AI system called “Lavender” to autonomously generate kill lists during its recent campaign in Gaza, dramatically transforming target identification and approval processes. Developed to swiftly process massive datasets, Lavender marked around 37,000 Palestinians as suspected Hamas or PIJ militants, including many low-ranking operatives previously untracked. The AI’s outputs were treated nearly as authoritative orders, with human officers often spending mere seconds verifying targets and rarely reviewing underlying intelligence, effectively sidelining traditional human judgment in lethal decision-making. This reliance on AI led to automated airstrikes, frequently executed against individuals in their family homes, resulting in high civilian casualties, including women and children. Lavender’s operation marks a significant shift in military AI use, highlighting both advancements and ethical concerns within the AI/ML community. The system rates individuals on a scale reflecting their likelihood of militant affiliation by analyzing pervasive surveillance data from Gaza’s population, enabling the military to manage vast numbers of targets in real-time—something infeasible for human analysts alone. However, with an error rate around 10%, the AI’s imprecision in distinguishing combatants from civilians has had grave humanitarian consequences. Moreover, the use of less precise “dumb” bombs against lower-ranked targets prioritized cost-efficiency over precision, further amplifying collateral damage. The unprecedented scale and automation of targeting underscore urgent debates on AI’s role in warfare, accountability, and the ethical limits of machine-directed violence.