AI Systems Used for Target Selection in Ukraine and Gaza Conflicts
AI systems have been deployed for automated target selection in military operations in Ukraine and Gaza, with systems analyzing vast surveillance data to identify targets and reducing human review time. An Israeli defense official acknowledged that this approach requires accepting margin of error, collateral damage, and risk of civilian deaths.
What happened
AI systems have been used for target selection in military conflicts. In the Ukraine war, AI analyzes satellite imagery, open-source data, unmanned aircraft recordings and field reports to identify targets for Ukraine's operations against Russian forces, with the time between detection and attack reduced to approximately 30 seconds.
In Gaza, Israel's 'Lavender' system identified approximately 37,000 potential targets based on visible connections between Palestinians and Hamas, according to reporting by the Guardian. Another system targets buildings, while a tool called 'Where is Daddy' tracks targeted individuals and sends signals to military units when they enter their homes for bombardment.
Reported impact
An Israeli intelligence official stated that using AI requires accepting a margin of error, acknowledging collateral damage risks, civilian deaths, and the possibility of mistaken attacks. Large volumes of surveillance and intelligence data are processed by technology with humans providing final authorization.
Index analysis
The systems reduce human oversight in the targeting process, with machine learning handling analysis of massive datasets and humans serving as final decision-makers rather than reviewers of individual targets. The compressed time between target identification and attack limits human review capability.
Governance controls
Human authorization remains the final step in attack decisions. However, the systems delegate target identification and much of the analytical process to algorithms, with human review occurring downstream after automated decisions have been made.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.