The Role of AI-Targeting Systems in Gaza and International Humanitarian Law

• The "Lavender" AI system identifies potential military targets in Gaza and Lebanon by analyzing vast amounts of personal and behavioral data.
• Researchers warn that the system systematically violates international humanitarian law by prioritizing quantity over quality and accepting high levels of civilian collateral damage.
• Critical infrastructure for this "kill-chain" is provided by major tech firms, including Google and Amazon through Project Nimbus, alongside software support from Palantir and others.

Mechanics of AI Target Selection

• Lavender assigns a risk score (1–100) to individuals in Gaza based on patterns of life, including mobile location, social structure, and surveillance data.
• A supplementary system, "The Gospel," automates the classification of buildings and infrastructure as military targets.
• The IDF has reportedly generated up to 1,000 potential targets per day, with thresholds adjusted dynamically based on available munitions and aircraft.

Legal and Ethical Violations

• The system design disregards core international law principles: distinction between civilians and combatants, proportionality, and precaution.
• The "Where's daddy?" protocol intentionally schedules strikes for nighttime when targets are at home with their families, maximizing civilian casualties.
• Human oversight is severely compromised by "automation bias," with supervisors often acting as mere "approval stamps" with only seconds to review AI-generated lists.

Calls for Accountability

• Analysts advocate for the immediate deactivation of such AI systems, greater public transparency, and an end to the export of automated targeting technologies.
• Legal experts suggest that the systematic use of these tools should be considered in ongoing international investigations regarding war crimes and potential genocide at bodies like the International Criminal Court.