AI-generated Key Takeaways
-
Starting December 8, 2020, Apple requires developers to disclose data collection practices of their apps, including those using ML Kit.
-
ML Kit for iOS collects device information, application details, performance metrics, and other data for diagnostics and usage analytics across all features.
-
Specific ML Kit features, like custom object detection or translation, may collect additional data such as model information or configured languages.
-
Some ML Kit features utilize Firebase services, requiring further data disclosure as outlined in Firebase documentation.
-
Developers should review the detailed data collection information provided for each ML Kit feature to ensure accurate and complete disclosure to Apple.
Apple requires developers publishing apps on the App Store to disclose certain information regarding their apps' data use. Apple has announced that these disclosures are required for new apps and app updates starting December 8, 2020. This document contains ML Kit library behaviors that could require disclosure according to Apple's guidelines.
This page lists the data collected by only the latest version of each ML Kit iOS SDK.
Collected data
Data collected in all features
| Data | By default, the ML Kit SDK collects... |
| Device information | Device information (such as manufacturer, model, OS version and build) and available ML hardware accelerators. Used for diagnostics and usage analytics. |
| Application information | Bundle ID and app versions. Used for diagnostics and usage analytics. |
| Device or other identifiers | Per-installation identifiers that are not intended to uniquely identify a user or physical device. Used for diagnostics and usage analytics. |
| Performance metrics | Performance metrics (such as latency). Used for diagnostics and usage analytics. |
| API configuration | API configuration (such as image format and resolution). Used for diagnostics and usage analytics. |
| Event type | Event type (such as feature initializations, model downloads, detection, resource releases). Used for diagnostics and usage analytics. |
| Error codes | Error code for feature events (such as feature initializations, model downloads, detection, resource releases). Used for diagnostics and usage analytics. |
Extra data collected in specific features
Custom object detection and image labeling
MLKitObjectDetectionCustom
MLKitImageLabelingCustom
- If you are using these libraries with custom models, the following data is
collected:
- Name, version, and URI/URL of the machine learning model for diagnostics
- If you are using
Firebase Model Deployment
with
MLKitLinkFirebaseto remotely host a custom model, refer to the Firebase ML model downloader and Firebase installations sections in Prepare for Apple Store data disclosure requirements.
Translate
MLKitTranslate
- Configured source and destination languages. Used for diagnostics and usage analytics.
- ML Kit uses Firebase remote config and Firebase installations for remote configuration diagnostics. Please refer to the Firebase remote config and Firebase installations sections in Prepare for Apple Store data disclosure requirements.
Entity extraction and digital ink recognition
MLKitEntityExtraction
MLKitDigitalInkRecognition
- Configured languages. Used for diagnostics and usage analytics.
Smart reply
MLKitSmartReply
- Identified languages. SmartReply internally uses language identification to only respond to specific languages. Used for diagnostics and usage analytics.
- ML Kit uses Firebase remote config and Firebase installations for remote configuration diagnostics. Please refer to the Firebase remote config and Firebase installations sections in Prepare for Apple Store data disclosure requirements.
Language ID
MLKitLanguageID
- Identified languages. Used for diagnostics and usage analytics.