The role of U.S.-made AI models in Israel's military campaigns has ignited global scrutiny amid surging civilian fatalities. Following Hamas' October 2023 attack, Israel reportedly increased its use of Microsoft and OpenAI technologies by 200x to accelerate target identification, per an Associated Press investigation. However, these systems have been linked to devastating collateral damage, raising urgent questions about Silicon Valley's responsibility in lethal AI applications.
Internal military documents reveal Israel stores 13.6 petabytes of surveillance data on Microsoft Azure – equivalent to 350 Library of Congress archives. This is the first confirmation commercial AI models directly enable warfare, warns Heidy Khlaaf, AI Now Institute’s chief scientist, in a
statement underscoring ethical crossroads for tech giants.
Key concerns include:
- Faulty machine translations misidentifying targets
- Biased algorithms flagging non-combatants
- Civilians comprising 70% of Gaza’s 50,000+ war deaths
Microsoft’s $133M defense contract with Israel enables AI-powered analysis of intercepted communications. While OpenAI prohibits weaponizing its models, revised terms now permit national security uses. Interviews with Israeli soldiers reveal rushed young officers relying on error-prone AI suggestions: We’re human – mistakes happen, admitted one reservist.
As Microsoft expands military cloud partnerships and Google alters ethics policies to accommodate defense contracts, critics argue unchecked AI deployment risks normalizing algorithmic warfare. For bereaved families like Mahmoud Chour, whose three daughters died in a disputed airstrike, the unanswered question remains: Why target a car filled with children’s laughter?