AI-Powered Targeting in Gaza: The Use of Lavender in the Israeli Conflict Raises Ethical Concerns

Image Credit: Jeff Kingma | Unsplash

Amid the conflict between Israel and Hamas, the Israeli military introduced Lavender, an AI-driven system designed to identify potential targets in Gaza. According to intelligence sources, Lavender identified tens of thousands of individuals linked to militant organizations like Hamas and Palestinian Islamic Jihad (PIJ). This unprecedented use of AI in warfare has raised serious ethical and legal questions, as well as concerns about the human toll of its application.

Lavender: The Israeli AI System and Its Role in Warfare

Lavender was developed by the Israel Defense Forces' (IDF) elite intelligence division, Unit 8200, which is comparable to the U.S. NSA or the U.K.'s GCHQ. The system uses machine learning to cross-reference data from multiple sources, providing intelligence personnel with a list of potential targets. In the initial stages of the conflict, Lavender identified as many as 37,000 Palestinian men as potential targets, based on their suspected connections to Hamas or PIJ.

This database of individuals marked as predominantly low-ranking members of Hamas was a significant factor in the military's operations. Intelligence officers working with Lavender described a shift from labor-intensive human assessments to an accelerated process where AI helped make critical decisions in identifying targets. While previously, target identification involved discussion and a legal review, the integration of Lavender allowed for a rapid and almost mechanical selection of potential operatives.

Evolving Guidelines: Collateral Damage and Targeting Policy

As the conflict continued, the IDF's approach to targeting shifted to accommodate the demands of an ongoing bombardment campaign. Intelligence officers reported that Lavender's algorithm was continuously refined during the first few weeks of the conflict, allowing it to achieve up to 90% accuracy in identifying individuals as Hamas or PIJ operatives. As a result, the system became a central tool in generating targets for the Israeli military, alongside another AI-based decision support system called Gospel, which identified buildings and structures for attacks.

However, a significant issue emerged regarding the acceptable number of civilian casualties. The IDF reportedly set pre-authorized limits on the number of civilians that could be killed in a strike aimed at a single Hamas operative, aiming to balance military objectives with minimizing civilian harm. These limits were set to expedite decision-making during high-pressure situations, but have raised questions about their alignment with international humanitarian law. Sources revealed that in some instances, up to 20 civilians were deemed acceptable collateral damage for targeting low-ranking militants, a figure that varied over the course of the conflict. This broad tolerance for civilian casualties led to widespread destruction and a shockingly high death toll.

Ethical Concerns: The Human Role and AI's Impact on Warfare

The use of AI in identifying human targets has brought ethical concerns to the forefront of discussions about modern warfare. Intelligence officers who used Lavender expressed doubts about the meaningfulness of their own involvement in the decision-making process, noting that their role often felt redundant. They believed their contribution added little value beyond validating the AI's decisions, which led to a sense of disconnection from the actual targeting decisions. Some felt that they were merely acting as a "stamp of approval" for the AI system's recommendations, raising questions about the role of human judgment in military operations driven by AI.

With Lavender generating tens of thousands of potential targets, officers also pointed to the military's preference for targeting militants in their homes, leading to increased civilian casualties. The focus was often on low-ranking operatives, and the use of unguided munitions, or "dumb bombs," meant that entire homes were destroyed, often killing all occupants. Despite having advanced technology capable of conducting surgical strikes, this strategy led to extensive collateral damage, further amplifying concerns about the morality of the IDF's actions.

Collateral Damage and the Human Toll

The consequences of the IDF's targeting strategies in Gaza have been severe. According to the health ministry in Gaza, over 33,000 Palestinians have lost their lives in the six-month conflict, and many more have been left homeless. UN data indicates that in the first month of the war alone, over 1,300 families experienced multiple losses, with 312 families losing more than 10 members. Such figures highlight the devastating impact of the IDF's targeting strategy, particularly given the leniency in collateral damage limits.

Experts in international humanitarian law have expressed alarm at the high number of acceptable civilian casualties set by the IDF, especially for lower-level targets. International law requires that proportionality be assessed for each individual strike, and the idea of pre-approved ratios for civilian casualties is seen as deeply troubling by many legal scholars. The IDF's official response, however, maintains that all operations were conducted in line with international law and that dumb bombs were used with precision to minimize civilian harm.

A New Era of AI in Warfare

Israel's use of Lavender and other AI systems in its war on Hamas has entered uncharted territory for advanced warfare, transforming the relationship between military personnel and machines. The reliance on AI-driven target identification raises fundamental ethical and legal questions about the role of technology in modern conflict and its impact on human decision-making. The testimonies of the intelligence officers involved reveal an unsettling reality: the human cost of AI-powered warfare is alarmingly high, and the implications for the future of military operations are deeply concerning.

The intense use of AI in this conflict highlights the urgent need for specific oversight mechanisms and ethical guidelines to govern the integration of machine learning in military decision-making. Without proper checks, the risks of repeating the tragic consequences seen in Gaza remain high. As advanced technology becomes more embedded in warfare, the balance between military efficiency and human morality must be carefully reconsidered to avoid the mistakes witnessed in Gaza.

Source: The Guardian