Apple AI privacy stays intact as Siri AI runs on Google Gemini servers

Apple reaffirmed its commitment to Apple AI privacy at WWDC, even as the newly‑branded “Siri AI” runs on Google’s Gemini models hosted on Nvidia‑powered Google servers.

Apple AI privacy promise amid Google partnership

The Cupertino giant explained that end‑to‑end encryption and on‑device preprocessing will keep user data out of sight from both Google and Apple staff. On‑device image analysis and local speech decoding remain unchanged, while the heavy‑weight language engine is off‑loaded to the cloud.

“Our privacy standards never waver, regardless of where compute occurs,” said an Apple spokesperson.

Analysts note that relying on external hardware spares Apple a ↑ $1.2 bn data‑center capex, but raises questions about jurisdictional exposure. The move mirrors a broader industry trend of hybrid compute models, as detailed by Reuters and Bloomberg. Critics argue that any cloud‑based processing, even with encryption, could be vulnerable to state‑level requests. Apple counters that its Private Cloud Compute architecture encrypts data in transit and at rest, and that de‑identification occurs before payload leaves the device. The company’s privacy narrative, long a market differentiator, now faces the test of transparency in a multi‑vendor ecosystem.

Words by Julian Reed (Consumer Electronics Expert).