Microsoft Israel head Alon Haimovich resigns after probe into Azure's role in military surveillance

Alon Haimovich has left his post as Microsoft Israel country general manager following an internal review into the unit’s work with Israel’s defense ministry and Azure infrastructure;

Alon Haimovich has stepped down as Microsoft Israel’s country general manager after an internal Microsoft investigation into the subsidiary’s work with Israel’s defense ministry. The personnel change coincided with departures among governance managers and a temporary transfer of oversight for Israel operations to Microsoft France, signaling an immediate company — level response to the probe and operational concerns.

Investigators concentrated on Azure infrastructure and the access arrangements provided to Israel’s Unit 8200. Leaked internal documents and reporting describe a “customised and segregated” area within Azure used to store vast volumes of phone — call recordings beginning in 2022, and reference roughly 11,500 terabytes of Israeli military data hosted on Azure servers in the Netherlands, with a smaller portion in Ireland. Microsoft said CEO Satya Nadella was not informed about the intended types of data stored there.

The review and actions reported by the company evolved through 2025. In May 2025 Microsoft said internal and external reviews found no evidence that Azure or Microsoft AI technologies were used to harm people in Gaza, while acknowledging it had provided software, professional services, Azure cloud services and Azure AI services to the defense ministry. In September 2025 Microsoft President Brad Smith wrote that the ongoing review supported parts of earlier reporting, and the company disabled certain cloud storage and AI services used by a unit within the Israeli defense ministry.

Reporting indicates the stored recordings were used operationally rather than merely archived. Sources say intelligence data were aggregated, transcribed and translated at scale — allegedly processing millions of mobile — phone recordings per day-and that the outputs were used during the Gaza offensive to prepare airstrikes and identify targets by cross — referencing with Israeli AI targeting systems. After public disclosures, parts of the dataset were reportedly moved out of the country and there were plans to transfer some processing to Amazon Web Services.

The episode highlights concrete governance and technical risks for cloud builders and their customers: segregated tenant environments, cross — border hosting (Netherlands, Ireland), limited visibility at headquarters, and AI-enabled processing can create compliance, export — control and contractual exposure across jurisdictions in Europe. Those architectural and contractual configurations make oversight and accountability difficult when services are repurposed for sensitive military uses.

For engineers, security teams and legal counsels the practical implications are clear: enforce strict access controls and tenant isolation, require and validate customer declarations about intended data types and uses, log AI-service provisioning and use, and codify data-residency and third — party processing restrictions in contracts and operational controls. These steps aim to reduce the risk that cloud or AI services will be used in ways that expose providers and customers to legal, contractual or ethical harms.