AI-driven targeting in Iran operation raises concerns over civilian harm
Reports indicate the United States and Israel used AI systems to rapidly identify targets during an offensive against Iran, with over 1,000 targets struck in the first 48 hours. International observers have raised concerns about the risks of algorithmic errors causing civilian casualties as AI accelerates military decision-making across all levels.
What happened
The United States and Israel reportedly used advanced AI systems to accelerate target identification during an offensive against Iran, according to US media reports. Over one thousand targets were attacked in the first 48 hours of the operation, including a chain bombing that killed Iranian Supreme Leader Ali Khamenei, an action that would have taken several days in previous conflicts.
Project Maven, the Pentagon's algorithmic warfare initiative now under the National Geospatial-Intelligence Agency, uses AI to analyze drone and satellite imagery, detect possible targets and mark them on maps. According to the Department of Defense, these algorithm-generated target lists are subsequently reviewed by human personnel. The Washington Post, citing three sources familiar with the system, reported the tool was used in the Iran operation.
Reported impact
Reports raise alarm about how these technologies are accelerating target selection and attack execution, and the risks that algorithmic errors could translate into civilian tragedies. AI is reportedly accelerating decision-making from target identification through legal approval to attack execution across all levels.
Index analysis
UN rapporteur Balakrishnan Rajagopal questioned the new forms in which wars are being fought and the use of AI platforms in the offensive against Iran. Similar systems have been tested in Syria, Iraq, Yemen, the Red Sea and more recently in Ukraine. Israeli military spokespeople have acknowledged using AI systems and advanced data analysis to help identify and prioritize targets in Gaza.
Governance controls
According to the Department of Defense, algorithm-generated target lists are reviewed by human personnel before execution. However, the accelerated pace of decision-making from target identification to attack execution raises questions about the adequacy of human oversight in the process.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.