An investigation by Israeli publication +972 Magazine and Local Call (via The Verge) reveals that Israel’s military has been employing artificial intelligence (AI) to select bombing targets in Gaza through a system known as Lavender, leading to substantial civilian casualties.
Developed in response to attacks on Oct. 7, Lavender at one point classified 37,000 Palestinians as suspected “Hamas militants” for potential assassination.
The Israel Defense Forces (IDF) denied having a kill list. However, it acknowledged the use of Lavender to CNN as "merely tools for analysts in the target identification process."
The spokesperson added that analysts "must conduct independent examinations, in which they verify that the identified targets meet the relevant definitions in accordance with international law and additional restrictions stipulated in IDF directives."
However, contrary to official statements, Israeli intelligence officers interviewed by +972 Magazine and Local Call claimed they acted as mere “rubber stamps” for Lavender’s selections, sometimes only verifying the target’s gender.
Lavender was fed data not just on known militants, but also on individuals with any loose affiliations to Hamas, leading to concerns about the criteria for classifying someone as a militant.
"I was bothered by the fact that when Lavender was trained, they used the term ‘Hamas operative' loosely, and included people who were civil defense workers in the training dataset," a source who worked with the data science team that trained Lavender told +972.
Also, Lavender was programmed to detect certain behaviors indicative of Hamas operatives, such as membership in a WhatsApp group with a known militant, regularly switching cellphones, or frequently moving from one address to another.
It operated on a scale of 1–100 to rank individuals’ similarity to known militants, with a changing threshold for targeting. Despite claims of 90% accuracy, errors occurred, including targeting individuals with similar names or connections to Hamas members.
Ethical And Accuracy Concerns With AI In Israel’s Military Targeting
"Mistakes were treated statistically," a source who used Lavender pointed to +972. "Because of the scope and magnitude, the protocol was that even if you don't know for sure that the machine is right, you know statistically that it's fine. So you go for it."
According to the report, Israeli intelligence officers had a permissive attitude towards civilian casualties. Initially, up to 20 civilian deaths were tolerated for lower-level targets, with “hundreds” acceptable for higher-ranking officials.
Another system, dubbed “Where’s Daddy?”, targeted individuals previously identified by Lavender in their homes. It was reportedly designed to target individuals who were at home at night with their families without verification of the target’s presence. See below.
"It happened to me many times that we attacked a house, but the person wasn't even home," one source revealed to +972. "The result is that you killed a family for no reason."
These practices raise concerns about the use of AI in conflict and the ethical implications of such technology, particularly regarding civilian casualties and the broader impact on Palestinian society.
Image credits: Anas-Mohammed on Shutterstock.
© 2024 Benzinga.com. Benzinga does not provide investment advice. All rights reserved.
Comments
Trade confidently with insights and alerts from analyst ratings, free reports and breaking news that affects the stocks you care about.
