Apple revealed that starting in March 2025, it will use the Apple Maps Look Around photos to train artificial intelligence. The tech firm also said it will use blurred imagery captured during Apple Maps surveys to improve other services.
Apple disclosed that it would use its Street View Maps photos to train generative AI models. The company revealed that the policy would go into effect in March 2025 and that it would maintain the blurred imagery when training AI.
Apple wants to use Apple Maps Look Around photos to train AI
Apple announced on its website that beginning in March 2025, it will use imagery and data collected during Look Around surveys to train AI. The company stated that it will use its Maps’ Look Around feature to “train models powering Apple products and services, including models related to image recognition, creation, and enhancement.”
Apple Maps Look Around cars will now also take photos to help train Apple Intelligence models pic.twitter.com/iB5OQ7MVg4
— Michael Nguyen (@MacMike1000) March 25, 2025
The tech company stated that it collects images and 3D data to enhance and improve Apple Maps using vehicles and backpacks (for pedestrian-only areas) equipped with cameras, sensors, and other equipment, including iPhones and iPads.
Apple also acknowledged that any images it captures, published in the Look Around feature, would remain blurred as part of its commitment to data privacy. It added that it would only use images with blurred-out details to train the AI models.
The company said it would accept requests from those who wanted their houses blurred. It clarified that images of houses collected during the surveys were not blurred out by default.
Apple’s image collection policy page lists dates and regions when it plans to collect new images for Look Around. People can access the policy page to identify when the company’s survey crew and vehicles will visit their region.
Apple currently offers features that rely on image generation models, such as Image Playground. The feature allows users of Apple Intelligence-compatible devices to write a prompt to create a new image.
Apple Intelligence also offers users Clean Up Photos, where they can remove objects from their favorite snapshots. Image Wand helps users improve their documents and notes by creating an accompanying image based on a rough line sketch using the Apple Pencil.
The Apple photo app also improved image recognition capabilities on Apple intelligence-capable devices. The feature enhances the accuracy of keyword searches and enables users to create memory-based movies via a prompt.
Apple launches Enhanced Visual Search on iOS 18.1 and macOS 15.1.
Last year, Apple launched a mechanism to identify landmarks and places of interest in images on the Photos application. The company said its mechanism would be available to its customers on iOS and macOS devices and enabled by default.
The feature sparked user controversy as software developer Jeff Johnson expressed concerns over Apple’s failure to explain the technology. The developer said the technology was available on iOS 18.1 and macOS 15.1. Apple later explained the technology, saying it was an enhanced visual search in photos that allowed users to search using landmarks or points of interest. It said devices privately matched places in pictures to a global index Apple maintained on its servers. The company added that it used homomorphic encryption and differential privacy to hide user IP addresses.
It explained that the encryption prevented companies from learning about the information in the photos. Apple added that users could turn off the enhanced visual search anytime on their iOS and iPadOS devices by accessing their settings.
The company later released a technical paper explaining the technology, published on October 24, when the Enhanced Visual Search debuted. It explained that the local machine-learning model analyzed the photos to look for a region of interest that may be depicted as a landmark. The company said that after the AI model found a likely match, it would calculate the vector embedding an array of numbers representing the image.
Google has used street-view images to train AI models for years; for instance, in 2017, researchers from the company used the images to teach a machine-learning model to create professional-looking photographs.
Cryptopolitan Academy: Want to grow your money in 2025? Learn how to do it with DeFi in our upcoming webclass. Save Your Spot
