How Israel uses AI to Hunt in Gaza

Scored, Tracked, Bombed: The AI Machine Accused of Targeting Gaza's Families An Israeli military artificial intelligence program known as "Lavender" is drawing renewed attention over allegations that it directed much of Israel's bombing campaign in Gaza. First revealed in April 2024 by a joint +972 Magazine and Local Call investigation, the system drew on mass surveillance data covering most of Gaza's 2.3 million residents, including phone contacts, social media connections and WhatsApp groups, scoring almost every person from 1 to 100 on how likely they were to belong to a militant group. At its peak, it marked around 37,000 Palestinians as suspected militants. Six Israeli intelligence officers told the outlets the army approved its kill lists for sweeping use despite knowing it was wrong in roughly 10 per cent of cases, sometimes flagging police, civil defence workers, or people who simply shared a name or phone with a suspect. One officer said he spent about 20 seconds on each target, mainly to confirm it was a man. Once a person was marked, a tracking system called "Where's Daddy?" alerted officers when they entered their family home, which sources said was deliberately treated as the easiest place to strike, usually at night. Junior militants were typically hit with unguided "dumb" bombs capable of flattening entire buildings, to save costly precision munitions. According to the sources, commanders permitted up to 15 or 20 civilian deaths per junior militant early in the war, and at times more than 100 to kill a single senior commander. UN figures cited in the investigation show that in the war's first month, 6,120 of those killed came from 1,340 families, many wiped out inside their homes. The Israeli military denies the core claims, saying its AI tools are only auxiliary aids and that an analyst examines every target individually. 🔔 Subscribe to our YouTube to stay up to date with our latest productions. 🎉 Introducing the official OnePath Network MERCHANDISE