Data AnalysisConsumer FacingVerified
Reviewed and published by trentmaziarz, March 22, 2026. Discovered and drafted by our automated research pipeline.
The Photos app on iPhone, iPad, and Mac uses AI to automatically scan a user's photo library, recognize faces, identify scenes and objects, and organize pictures into searchable albums — all without sending images to Apple's servers. Users can search for "dog at the beach" or "birthday cake" and find matching photos from years ago. These features have been present since iOS 10 (2016) and have been continuously improved.
Details
Apple uses machine learning models running on the Apple Neural Engine — a dedicated processor chip inside Apple devices built specifically for fast AI computations — to analyze photos on-device. Face recognition groups pictures of the same person together into a "People & Pets" album (pets were added in iOS 17 in 2023). Scene analysis powers natural-language photo search and the automatic creation of Memories videos from related images. Apple's own machine learning research has been published on the technical details, describing a pipeline that generates embeddings (mathematical summaries of faces and objects) to match people across thousands of photos. An additional AI-powered feature called Clean Up, part of Apple Intelligence, was added with iOS 18.1 to let users remove distracting objects from photos.
Products affected
iPhone (iOS 10+)iPad (iPadOS)Mac (macOS)Photos app
Sources & Evidence
Company Disclosure
Other practices by Apple
OtherSafari's Intelligent Tracking Prevention (ITP) uses machine learning running directly in the browser to identify advertising networks and data companies that track users across different websites, and then blocks or limits their ability to do so. First introduced in September 2017, ITP has been updated continuously and now blocks third-party tracking cookies entirely by default.Creative GenPersonal Voice is an accessibility feature introduced with iOS 17 in September 2023 that lets users create a synthetic voice that sounds like their own. It is designed primarily for people who are at risk of losing their ability to speak due to conditions like ALS. After recording 150 short phrases (roughly 15 minutes of audio), the device processes the recordings overnight using on-device machine learning and creates a voice clone the user can speak through in calls, FaceTime, and apps.OtherApple Watch uses machine learning algorithms to detect when a wearer has had a hard fall or been in a car crash, and can automatically contact emergency services if the person is unresponsive. Fall Detection has been available since Apple Watch Series 4 in 2018. Crash Detection for severe car accidents was added with the iPhone 14 and Apple Watch Series 8 in 2022.
Have evidence about Apple's AI practices? Submit a report.
Report a Sighting →