Apple tests privacy-focused Siri AI in mass public beta
Apple has released a public beta of its overhauled, AI-powered Siri, kicking off the largest consumer test of a voice assistant built to rival ChatGPT while processing user data locally to navigate strict privacy rules.
Apple is releasing the iOS 27 public beta, giving everyday users their first chance to try a completely redesigned, AI-powered Siri. The update transforms the company's aging voice assistant into a tool capable of reading emails, photos, and messages, as well as answering general knowledge questions.
With 2.5 billion active devices worldwide, even a small adoption rate makes this the largest real-world test of Apple's answer to ChatGPT, Gemini, and Claude. The upgraded assistant is available across the iPhone, iPad, Mac, Apple Watch, CarPlay, AirPods, Apple TV, and Vision Pro.
The technical architecture is what will matter most to European businesses and regulators. Rather than relying entirely on remote servers, Siri uses Apple's Foundation Models that run directly on the device, supplemented by Private Cloud Compute, which guarantees that personal data is neither stored nor accessible to Apple. Apple built these models in collaboration with Google, using a process called distillation to turn Google’s Gemini into smaller, highly efficient models tailored for Apple Silicon.
This on-device approach is a significant market differentiator in Europe, where enterprise adoption of generative AI has been cautious due to data sovereignty concerns and strict GDPR enforcement. By keeping AI processing local, Apple is offering a privacy-compliant alternative to cloud-heavy competitors, potentially accelerating corporate uptake of its hardware.
In daily use, Siri is now deeply woven into the operating system. Users can trigger it by voice, side button, or by swiping down from the Dynamic Island, and it is integrated into the Spotlight search tool. Apple has also given Siri a standalone app, though its deep system integration makes a separate application somewhat redundant.
Early tests of the developer version show the assistant successfully handling tasks like summarising group texts, adding calendar appointments from messages, and identifying nutritional information through the camera. However, the software remains imperfect. It occasionally throws error messages or fundamentally misinterprets queries, such as searching a user's contacts for the word "Iran" when asked for international news.
Despite these bugs, the developer betas have been relatively stable this year. Consumers who rely on their phones for critical, error-free tasks should still wait for the official September launch, but the beta marks a clear step in Apple's strategy to embed AI into daily life without sacrificing user privacy.
