Microsoft blocks Israel's use of its tech. in mass surveillance of Palestinians (www.theguardian.com)

Microsoft has cut off an Israeli military intelligence unit’s access to some Azure cloud and AI services after a Guardian-led investigation revealed the agency used Microsoft infrastructure to store and process a massive repository of intercepted Palestinian cellphone calls. The company says Unit 8200 violated its terms by holding content-level communications in a segregated Azure environment that enabled AI-driven analysis and long-term retention; the cache reportedly reached as much as 8,000 TB and was described internally as processing “a million calls an hour.” The decision follows an urgent external review launched after reporting and employee/investor pressure, and Microsoft says it has “ceased and disabled a set of services” to the unit while continuing an ongoing inquiry. For the AI/ML community the case underscores how commercial cloud scale, storage and ML tooling can be repurposed for indiscriminate mass surveillance and targeting. Technically, the story highlights risks around third-party hosted sensitive intelligence—near‑limitless compute + segregated storage + AI analytics enabled content playback, indexing and target research—and shows governance gaps in vendor oversight, employee transparency, and rules-of-use enforcement. The episode may prompt tighter vendor policies, stricter export/hosting controls, and stronger audits of how cloud and AI platforms are used by state actors.