Microsoft and OpenAI AI infrastructure used in Gaza and Lebanon operations
According to Associated Press investigations, the Israeli military has used AI infrastructure from Microsoft and OpenAI for targeting operations in Gaza and Lebanon since October 2023. The reporting documents cases where AI systems misidentified targets, including high school students classified as potential combatants, resulting in civilian casualties.
What happened
According to Associated Press reporting, the Israeli military has extensively used AI infrastructure from Microsoft and OpenAI since October 7, 2023, with usage increasing 200-fold, including via Microsoft Azure cloud services for mass surveillance data collection and analysis of communications.
The reporting indicates AI systems were used for targeting civilians in populated areas of Gaza and Lebanon. A $133 million contract between Microsoft and Israel's Ministry of Defence was signed in 2021, making Israel Microsoft's second-largest military customer after the United States.
Reported impact
According to an Israeli intelligence officer cited in the reporting, AI targeting errors have occurred due to misinterpretation and misidentification, with one example being high school students misclassified as potential combatants.
Index analysis
The reporting states that when AI systems are combined with human decision-making, accidental killings and increased civilian casualties result. Former OpenAI manager Heidi Khalaf is quoted as saying this is the first confirmation that commercial AI models are directly used in warfare.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.