How Siri Finds Driver's License Info in iOS 18

In iOS 18, Apple Intelligence grants Siri deep personal context, allowing the assistant to locate specific personal information, such as a driver's license number, directly from your stored photos. By combining on-device computer vision, semantic indexing, and large language models, Siri can understand natural language requests, identify relevant documents within your photo library, and extract the precise alphanumeric data you need without compromising your privacy.

On-Device Semantic Indexing

The foundation of this capability is Apple Intelligence's semantic index. iOS 18 continuously organizes your personal data locally, creating a map of relationships between your messages, emails, calendar events, and photos. Instead of merely scanning filenames or user-added tags, the semantic index understands the contextual meaning of your content, cataloging images that resemble government-issued identification cards, receipts, or medical forms.

Computer Vision and Optical Character Recognition

To read the details on a physical card, Siri leverages the system's built-in Vision framework and advanced Optical Character Recognition (OCR), evolved from Apple's Live Text technology. When a photo of a driver's license is added to your library, on-device machine learning models process the visual layout:

  1. Document Classification: The model recognizes the visual structure, layout, and iconography typical of an official ID card.
  2. Text Detection and Extraction: The OCR engine isolates the text fields, identifying labels such as "DL," "DOB," "EXP," and the corresponding identification strings.
  3. Entity Recognition: The system categorizes specific text blocks as personal identification numbers rather than random text.

Natural Language Processing and App Intents

When you ask Siri, "What is my driver's license number?", iOS 18 uses on-device language models to parse the intent of the prompt. Siri recognizes that the request requires retrieving a specific credential belonging to you.

Through the App Intents framework, Siri queries the Photos database via the semantic index, retrieves the image flagged as your driver's license, extracts the specific string associated with the license number, and presents it directly in the interface. Siri can also display the original photo as verification.

Privacy and Local Processing

All scanning, indexing, and text extraction for stored photos take place directly on the device using the Apple Neural Engine. Your photos and personal identification numbers are not transmitted to external servers to perform these searches. This ensures that sensitive documents stored in your photo library remain private and inaccessible to third parties while still being instantly retrievable by Siri.