Overview
- Bloomberg’s Mark Gurman reports some employees testing iOS 26.4 find the new Siri inconsistent and unreliable on basic tasks.
- Apple is running an internal comparison between on‑device AI models and a cloud‑assisted approach using Google’s Gemini via Private Cloud Compute.
- The Siri overhaul was postponed after failing internal quality standards, with Craig Federighi saying it required rearchitecture for a 2026 debut.
- Ongoing turnover in Apple’s AI ranks, including departures from the Foundation Models and AKI groups, raises continuity and resourcing concerns.
- With several months before the expected March–April window, Apple can still address issues, though further delays or a shift in approach remains possible.